You are on page 1of 44

ROMANIA GOTHICA II

The Frontier World


Romans, Barbarians and Military Culture
Proceedings of the lnternational Conference at the Eofoos Lordnd Uniaersity,
Budapest, 1--2 October 2010

TTaDAR Vma ("d.)

with assistance from


Puu-rP RANCE
(English language editor)
and
Apnrux Blay, IstvAN Koxcz, LrvENTE Seuu

Edtvos Lor 6tnd University


Institute for Archaeological Sciences

Martin Optrz Kiad6


Budapest/ 2015

quam gentilitate app ellant.

THE PHILOLOGICAL EVIDENCE FOR GERMANI


IN THE LATE ROMAN ARMY:
Germanic Loanwords in Roman Military Vocabular5l
Philip Rance

Thelpurpose of this paper is to examine the number, etymology and chronology of Germanic loanwords in Roman military vocabulary with a view to establishing what these lexical influences can
reveal about the character of intercultural contact in this sphere during Late Antiquity. From the beginning of the twentieth century, philologists systematically assembled examples of such loans, usually
in general surveys of linguistic interaction between Romans and Germani, and their interpretations
have to varying degrees penetrated selected entries in Latin and Romance etymological lexica.2 More
recently, individual case studies have elucidated the documentation and dynamics of Germanic loanwords in Roman military culture, and exposed some errors and misconceptions in older scholarship.3
Nevertheless, a comprehensive list or discussion is lacking, and the bibliography remains widely scattered across diverse periodicals of distantly related disciplines. While loanwords have long been recognised as a fertile source of cultural informatiorL the purpose here is to chart the linguistic impact of
Germani within one particular Roman institution. Most of the linguistic borrowings examined in this
study occurred originally in Vulgar Latin, and in particular the demotic idiom spoken in the army the elusive sermo castrensis, a fusion of official nomenclature, technical terminology and soldiers' slang.
In this linguistic register we can dispense with any notion of loanwords motivated by social or cultural
cache! although, as in modern armies, the use of a common argot or idiolect may itself be expressive
of shared community and professionalism.a
Consequently, some genres and categories of source material are more useful than others as witnesses to Germanic loanwords in this sphere. The main narrative accounts of Roman military operations in Late Antiquity are classicizing histories, whose authors adhered to the stylisiic conventions
of Kunstprosa and endeavoured to avoid termini technici and foreign usages (which for Greek authors
included Latinisms) by formulating periphrastic expressions or seeking analogous vocabulary precedented in classical historiography. Only in rare cases do classicizing historians employ such technicalia, typically couched in circumlocutory or epexegetical phrases which aim both to explain and
to excuse a particular usage as a term commonly employed by contemporary Roman soldiers and/or
barbarians.s In the sixth century, chronicle sources, less concerned with literary proprieties, can proI am grateful to Tivadar Vida for his kind invitation to participate in a stimulating conference. This paper was written and
presented during the course of a Humboldt-Forschungsstipendium fiir erfahrene Wissenschaftler, hosted by the Institut
fiir Byzantinistik, LudwigMaximilians-Universitiit, Miinchen, 2009-11. Abbreviations of ancient authors and works follow
the conventions of The Oxford Classical Dictionary (19963) and Ann6e philologique. Unless otherwise stated, all translations from Latin and Greek are my own.
Krurr 1901,386-7; Hrnarus 1902,263-5; BniicH 1913, 15-77,79; Geurrr-scnnc 7970,78-19,22-3.See more recently KorIas 1993,
40-41; GnnoN 1998,184-6,189-94; R.tNce 2004b 305-10. KuuN 1972 is largely concerned with Latin influence on Germanic
vocabulary, but briefly discusses reverse currents at 43-5, 53.
E.g.

Knaurn

1987; 1996; 1997; RaNcn 2004a; 2004b; 2014.

Attempts to define the nature of sermo castrensis and to trace its constituent features and vocabulary include KeN{rr 1901,
who originally coined the phrase; Henaeus 1902; RrrcnrNxnoN 1961; MrH.i.nscv 7968-9;1969;1977;7978-9; Nacsr 1974,118-20;
Moscr Sassr 7983;Pfxnz Casrno 2005. See also remarks on'the pidgin of the Mediterranean armies'in the sixth century by
Alvronv 1996, 102-8.
E.g. Procop. BV 2.2.1,, 10.4; Theoph. Sim. Hisf. 2.4.7; 3.4.4; 6.9.1.4. See general remarks
urnorv 1964,326; Cauunou 1970,78-9; WuIrrv 1988,334.

in Zlrrracus

1935, 732-3;

CaruttoN-Ce-

51

PHILIP RANCE

vide a glimpse of the'real-world'vocabulary of the late Roman army. Similarly, documentary papyri
generated by the army are also valuable witnesses to technical usages, even if they usually relate to a
limited range of routine bureaucratic procedures. Our most important sources are military treatises,
notably Vegetius' Epitoma rei militaris (383-450) and Maurice's Strategicon (completed late 590s).6 Maurice
is of particular value insofar as he expressly eschewed stylistic and literary considerations in favour of
a simple Koine Greek comprehensible to army officers and replete with the Latin and semi-barbarised
technical jargon of their profession.T The evidence of this treatise was almost entirely overlooked in
older studies of this subject.
The nature of the source material poses problems of chronology. In several instances it is safe to
assume that the first appearance of a Germanic loanword in a text, whether or not the author has literary objectives, is no secure indicator of the date of the loanwordb adoptiorL the extent of its currency
or the period of its usage. Parallels may be drawn wittr, for example, certain Late Latin military and
equestrian terms derived from Gaulish that are first attested from the late fourth century (e-g. drungus,
bebra, gulbia), but which must have enjoyed a'prehistory' in Vulgar Latiry perhaps for several centurieg
before their emergence in a written source, even if we accept the survival of Gaulish as a substrate language into the fourth century.s Correspondingly, some Germanic loanwords that entered Latin in the
fourth and fifth centuries might first be documented only in sixth-century or even later sources, and
in most cases the occasion and circumstances of the original borrowing remain imprecise or a matter
of guesswork. An important consequence of this time lag in the documentation is that a Germanic loan
into Latin military vocabulary may not be directly attested in Latin sources but only in transliteration
or as a secondary loan in later Greek texts produced in the East Roman Empire. This pattern of evidence reflects the fact that Latin continued to be the official language of command, instructiory exhortation and discipline in the East Roman army well into the seventh century, long after the demise of its
counterpart in western Europe, and even after a functional proficiency had ceased to be a requirement
in other institutions and professions of the East Roman state.e
Identification of Germanic loanwords is based on several criteria. Explicit statements by Roman
writers might indicate that a term is Germanic or at least'barbarian' in the context of Germani, as
exemplified by the quotation of Ammianus Marcellinus in the title of this paper: 'which in the foreign
tongue they name ...i or elsewhere, 'which the barbarians call ...'.10 Such testimony is not without complications: the author's opinion may be faulty, while his silence or ambiguity with regard to the identity
of 'they' allows uncertainties to linger. With or (more usually) without such assertions, comparative
philology can posit a Germanic etymon of a Latin (or Latin-derived Greek) term on the basis of correspondents in one or more Germanic languages, although some of the examples examined below reveal etymologies based on slender evidence or superficial analogies. I have excluded from the present
study Germanic loanwords attested only in Romance (primarily Gallo- and Ibero-Romance). Older
scholarship, notably a monograph by Brtich (1913), attempted to define chronological criteria whereby
the Romance evidence could be interpreted as indirect testimony to the currency of a Germanic loan
in Vulgar Latin prior to c.AD 40O on the assumption that the occurrence of cognate terms in two or
more western Romance languages is indicative of a Germanic loan into their common ancestor before
political and cultural fragmentation of the western Roman Empire intensified regional linguistic divergence. Without discounting the possibility that this may have been the case with some specimens of
military vocabulary, this approach rests on too many imponderables and historically arbitrary judge6
7
8
e

Editions: Rrnvn 2004; DrNNrs 1981. Vegetilts' Epitoma was written at a much-disputed point between 383-45Q with a date in
the reign of Theodosius I (379-95) favoured here. The evidence and arguments are most recently assessed by Charles 2007,
who prefers a fifth-century date.
Maur. Strat. pr.27-31: ...60ev rcai ?copaircaig noA),axg rcal &AAatg iv otga,rLr.r,rLrr1 ouvqOeiq tetqrrppvatq Xqrlpe0a
AtEeot,'... to which end we have also frequently employed Latin and other terms which have been in common military
usage ...' (29-30).
RaNcn 2004a,102-5.
See most recently Rar.rce 201Q with the bibliography cited at 64 nn. 'L-2.

10 Seebelown.121.

52

THE PHILOLOGICAL EVIDENCE FOR GERMANI IN THE LATE ROMAN ARMY

ments to command confidence, not only on account of the complexities of individual examples, but
also given that loanword traffic between western Romance languages in relation to military dress,
equipment and practices is demonstrable at a later date, particularly in connection with the expansion
of the Frankish empire and the diffusion of feudal institutions. Certainly some of the pre-400 military
contexts hypothesised to account for Germanic terms in western Romance now seem strained and/or
anachronistic, while the absence of correspondents in Balkan Romance or the Latin-based argot of the
East Roman army, even allowing for a relative isolation or distinctiveness of Balkan Latin prior to c.400,
points to later regional developments and calls into question whether the western Romance evidence
could reflect elements of an'institutional language' of the late Roman army.11
Relatively few terms directly attested in Roman military vocabulary have been assigned a Germanic derivatiory and closer scrutiny both reduces their number and casts doubt on the Germanic
identity of some of those that remain. Questions relating to the overall character and motivation of
these loanwords, as well as an attempt at periodizatiory are left till the end, but it will be useful here
to distinguish several broad categories. First, as with any multi-ethnic imperial army, the adoption of
a foreign garment, item of kit or practice by Roman soldiers might entail the simultaneous adoption of
its foreign name; a straightforward case of loanword for loan-object - borrowed names for borrowed
things.12 In general such loans tend to be nouns and relate to technological innovations and improvements or decorative novelties. Second, in contrast to these linguistic expressions of cultural transfer,
other Germanic loanwords reflect a nominal borrowing only. In some instances, a new Germanicderived word appears to have been applied to an existing or broadly corresponding Roman item or
practice within the context of a shared military culture. In these circumstances, one must take account
of potential Roman misunderstanding of barbarian languages and cultural concepts which might have
led to an inexact equation between the original meaning of the Germanic etymon and the usage and
evolution of its derivative in Latin and/or Greek. In some cases the dynamic of the loan may have been
promoted or affected by confusion with or assimilation to a similar-sounding Latin or Greek word
(e.g. barditus-barritus, burgus-nuqyoq). Within this category, a distinction must be drawn with instances
where a Latin author merely reports a term in use among the Germani, which may then gain currency
in literary Latin as a non-technical usage, but there is never any question that this loanword was ever
part of Roman sermo castrensls, still less reflects actual technological transfer (e.g. Tacitus onframea). A
third category is a hybrid of the first two, where a Germanic etymology is secure but the loan may be
the product of either process, reflecting a degree of uncertainty with regard to the specific or generic
usage of the term in Latin. Fourth, older scholarship assigned a Germanic etymology to several late
Roman military terms which, in the present state of the evidence, must be considered uncertain or
doubtful. To some extent these attributions reflect the predominance of German scholars in the fields
of classical and comparative philology in the nineteenth and early twentieth centuries. Finally, as the
proven'rejects' of the preceding category, we must acknowledge and discard those Latin or Greek
words which have had a Germanic derivation thrust upon them that is now demonstrably erroneous.

Bniicu 1913, with explanation of principles at 2-4, 7927, 48-9, and examples of Germanic military terminology in Romance at esp. 61-2, 99-105, 142-3, with in some cases rather imaginative interpretive evocations of soldiers' lives. See also
Gaurr.r-scnnc 1970, 279-81,, 286-7, 318, 337, 378. Briich's line of argument and examples are rehearsed by GnarN 1998, 183-4,
189-95, but with recognition of the possibilities of loanword traffic between Western Romance languages arising from
Frankish expa nsion: 19 4, 197 -8.
I have here borrowed the phrase of Wno 1970, who examines the cultural dynamics of Latin loans into Insular Celtic.
See

53

PHILIP RANCE

Cerscony 1: Bonnowrn Nalrss ron Bonnowso THrNcs:


tufa, roOQa, rouQiov
Late Latin tufa is first attested in relation to an ill-defined type of standard or emblem, which the Roman army had apparently adopted as an additional distinctive device on the battlefield. Later tufa and
rouQa, with diminutive touQlov, also became a generic designation for various forms of ornamentation made of animal hair, such as plumes, crests and tassels. The only pre-medieval witness to this
term in Latin is Vegetius, who includes tufa in a list of the different types of flag or standard used to
issue visual signals in combat muta signa sunt aquilae dracones aexilla flammulae tufae pinnae, 'the "mute"
signals are eagles, dragons, ensigns, flammulfle, tufae, plumes'.r3 Vegetius has previously introduced or
defined the preceding four Latin items.ra The lack of explanation concerning the form or origin of tufae,
clearly a foreign term, is therefore unusual and may indicate that this word already enjoyed a wider
currency. It is not impossible that the following wordpinnae is itself a gloss on tufae - certainly the later
evidence indicates that tufa could signify a type of 'plumei though this arrangement is not typical of
Vegetius'method of terminological clarification nor do the manuscripts give any reason to suspect that
pinnae is a later editorial insertion.ls
In the mid sixth century, two writers of very different genres provide further information. Around
the 540s, the Christian cosmographer Cosmas Indicopleustes mentions a species of large long-haired
'wild ox' (ayqLopouq) native to India clearly identifiable as the wild yak, which locals reportedly pursue in order to obtain its tail-hair. Cosmas observes that this animal is the source'from which comes
the so-called tufa (ro0Qa), with which officers adorn their horses and standards in the field'.16 It is
unclear whether he refers to the Roman army or indigenous Indian peoples, but the wider evidence
favours the former interpretation (see below). Cosmas does not necessarily imply that tufa is synonymous with or exclusive to the hair of this remote and exotic beast, only that yak hair is one example of
the type of animal hair that Romans might use for decorative purposes and call tufa. Ttre bureaucrat
and antiquary John Lydus, probably writing in the 550s-560s, provides a more explicit definition of tufa
in his attempt to reconstruct the regalia of the kings of ancient Rome. Among their royal insignia he
includes: 'long spears, likewise twelve in number, which have not bladed tips but crests (AoQLag) suspended from them, which the Romans call iubae, but the barbarians tufae (rouQag), having corrupting
the word somewhat'.17 These tufae are distinct from the aexillawhich follow in the list. Lydus appears to
complement the meagre data supplied by Vegetius: a tufa is a specific type of crest or emblem mounted
on a shaft. The word AoQu denotes the hair of an animaf whether the mane of a horse or birse of a
boar in situ, or animal-hair tufts employed as plumes or other ornamentation. Lydus identifies tufa as a
contemporary word of barbarian origin, which he construes as a phonological corruption of the Latin
iubfl,'rnane, hair, cresti a piece of spurious etymologising that may safely be set aside. A few decades
later, Maurice employs touQ[ov (syncopated variant rouQlv), a diminutive of to0Qa, to denote both a
plume or crest on a helmet and a decorative attachment to a harness, presumed to be an animal-hair
tassel, and thus corresponding directly to the earlier remark of Cosmas Indicopleustes.l8 Such tassels affixed to harnesses and bridles are depicted in near-contemf/orary representations of horsemen
and mounted warriors, and were clearly part of a common martial culture shared with neighbouring

"la

V"9.Epit.3.5.8.
Generally Ve g. Epit. 2.73; also aquila: 2.6.2, 7.3; draco: 7.23.2; 2.7.5; flammulae: 2.1,.2.
ls Vegetius otherwise uses pinnae only at 3.8.72, where it has an alternative
meaning of 'battlements'.
16 Cosmas Indic. 11.5: . . . i!, o0 Atyopivq tonQa, rcoopoOor touq lnnouq rcai. ta pav6a ol dqlovteg eiq touq rcapnouq.
{
1
n l. Lyd. De mag. 1.8: 66gata intpr1rcr1, ,ooaur,og tov aqt0pov 6uorcalberca, arcqo[rQibaq plv o0r ircovta, r,loqeprdvag 6i
AoQLriq, rcaAo0ot 6i autag oi ptv 'Pt rpaioL ioupaq, oi bi priqpagoL toriQaq, pqalu tL naqaQ0aqe[or1q trlq Adeors (Bandi
-18.27-3);
cf. capitulatio $4: 6ta t[ tag AoQraq tweq toJQaq raAoOow, 'why some call crests tufae' (Bandi 4.15).
18 Maur. Strat.7.2.12;12.8.4.3 (plume or crest on a helmet); 1.2.23-4,44-5 (ornamentation on a harness). The case for an animalhair tassel is supported by the gloss rltoL tqilaq, br hair', inserted into the tenth-century Ambrosian paraphrase (1.2.12),
see

54

Lrour 2003, 33.5.

THE PHILOLOGICAL EVIDENCE FOR GERMANI IN THE LATE ROMAN ARMY

peoples.le The term touQlov retains this meaning in derivative tenth-century Byzantine military treatises.zo From the tenth century, toOQa is also applied to a particular type of imperial headdress, first

documented (without this terminology) under Justiniary which was worn during victory celebrations
and comprised an elaborate panache of feathers.2l
John Lydus identified tufa as a'barbarian' term. A Germanic cognate was signalled as early as the
eighth century, when Bede (c.731) relates that Edwir; King of Northumbria (r.616-632/3), exhibited his
overlordship by having carried before hirn illud genus oexilli, quod Romani tufam, Angli oero appellant tuuf,
'that sort of banner, which the Romans call tufa and the English tuuf'.22 Bede was demonstrably familiar with Vegetius' Epitoma, which he elsewhere quotes or paraphrases without acknowledgement.23
As the unique ancient witness to the Latin term, the Epitoma was probably Bede's source also in this
instance. Bede therefore draws an erudite etymological comparison between Latin and contemporary
vernacular. The Old English term to which Bede refers is the masculine noun pilf,'tuft,banner', which
is independently attested in Old English literature. Even if the practice Bede describes is an Anglian
attempt at imitatio imperii, there is no possibility that pilf is a Latin loanword and not an organic element within Old English (see adj. pilfe, 'tufted, bushyl pilfig, gepilf, 'leafy'; noun pift, ,thicket,; p1ifel,
'busll, thicket, foliage'.2a The Germanic derivation of Late Latin tufa is universally accepted in modern
scholarship.2s It remains puzzlingthat no direct correspondent can be identified in any other Germanic
language.26 Attempts to posit a Germanic etymon of tufa have tentatively offered a Gothic masculine
noun*pilfaz, although the Old English evidence complicates attempts to narrow the possibilities to an
East Germanic derivatiory even if this might be more attractive on historical grounds.2T
The historical and archaeological evidence provides no specific information concerning the cultural background to or circumstances of this loar; but by implication the Romans adopted an animal-hair
standard from a Germanic people(s) at some point before the later fourth century. For comparative purPoses, broader ethnic and geographical horizons are potentially instructive. For example, both Sasanian and Steppe parallels have been adduced for the use of decorative tassels on equestrian equipment,
though the complexity of cultural currents across Central Asia hinders definitive answers.2s More intriguingly, Cosmas' specification of yak-hair tassels hung on officers'banners, together with the John
Lydus' description of an animal-hair crest on a long pole, are strongly reminiscent of a type of emblem
or standard, comprising one or more horse- or yak-tails mounted on a pole, which is documented in
different periods among various martial nomadic pastoralists of the Steppe. Later Turkic languages

t'
'o
21
22
23

2a

"

26

27
28

E.g. an equestrian portrait of St. Menas in a sixth-/eighth-century Coptic manuscript (John Rylands S.33), see Stephenson-Dixon 2002,61; and the (Turkic?) armoured horseman depicted on vase 2 of the sixth-/tenth-century Nagyszentmikl6s
Treasure, see Alfoldi 7950,124-38, pls. I-IL
Leo, Tactica 6.2-3,1Q 21 (Dennis); Sylloge tacticorum 38.5, 39.1,,3 (Dain).
Pttrz 7977, 49-50,57-8; McConur cx 7986, 148-9, 728,417-18; SranurusoN 2003, 56-9.

Bede,HE2.16.
Jorvns 1932; RnrvE 2004, xiii-xiv, xxviii.
Boswonrn-Torr.rn 1075, s.vv. pif, h,ifr, b,lfig,

S:: Martin

prift, with

457, s.v.

gepilf,

874, s:r. sigepilf (?), 1084,

s:t. pyfel.

Kempf 1901, 348,368,387; Heraeus 7902,264; Briich 1913, L6, 45, 1"18; Walde-Hofmann]J7l1', s:r. tufa;
Mihdescu 1968-9,1488-9 ('probablement');7978,32,305 (5922,309); Kuhn 1972,49,46; Gamillscheg19Z0,71-l9;Green 1998,
234. Schramm 7954, 1 248-9 accepts the Germanic derivation of Latin tufa, but is eccentric in denying a connection with
Old English adjectival and nominal parallels in relation to foliage and insisting on a primary ."*u.rti. correspondence
to Old Norse pifa, whichhe understands as 'Grabhiigel', although in fact a generic 'mound'; see Cleasby 19SZ,7SO, ss. ptifa.
1892;

Ernout-Meil1et706, stt. tufa citeBedebut do not commit to an etymology.


The origin of the personal name of Tufa, a late fifth-centwy magister militum is disputed. See older literature in Wnaon 1g91,
121, n.5. ManrrN 1892 posited a connection between Tufa, construed as a nickname 'Helmbusch', and Late
Latin tufa.This
view is endorsed by ScnournLo 1911,242-3; followed by PLRE 1131: Tufa. In contrast, R. K<igef Review of Wnxrn 1891 in
ZfdA (N 'F. 24) 36 (1892), Anzeiger (1 Jan. 1892) 43-60 at56 makes a more persuasive case that'T6fa ist ein gut deutscher und
wahnscheinlich sehr alter name', unrelated to Latin tufa or Old,Englishpilf.
Gamillscheg7970,ll253,257; Kobler 1989, 253, sst. "pu-f-az (< East German ic puba-). See also Ktthn1972, 46, whoposits Germanic*pifo or *pubo. see broader Indo-European parallels in pokorny 1959, 1090-g1.
Kondakov 7924;861int1978,786-9,204-7 with 193 figs. 10.4,70.10; B6na 198Q 55-62; Szddeczky-Kanooss 1987, 6;;Fuxar- Honrucnr 1969-84, If pls. xxxv, xlii, xlviii-lii; lY 77-3; Kouas 1988, 61.-4;Dl.ru 2003, 566-7.

55

PHILIPRANCE

designate such a standard as a tu! (< Chinese tu, 'ban'rter, standard').2e While any attempt to connect
Late Latin tufa and its posited Germanic antecedent with Turkic tulwould face insurmountable philological and chronological obstacles,3o this oriental perspective draws attention to the appeal of certain
emblems, crests and totems across and between different and distant cultures. The fact that some East
Germanic peoples, including the Goths, had Steppe forebears and/or spent long periods under Hunnic
hegemony would have facilitated cultural exchange in this sphere.
carrago

Late Latin feminine carrago designates a wagon-train, providing transport and originally habitation,
which can also serve as a defensive barricade or laager. The term is first attested in late fourth-century authors, though awareness of the phenomenorL without the terminology, can be traced in Roman
sources back to the mid third century. According to Ammianus the word is Gothic.31 Scholars have

posited Gothic feminine noun*carr(h)dgo,'wagon-hedge' (cf. Germanic*hag- > Old High Germanhag,
'hedge, fence'), adopted into Latin by analogy to a feminine declension -ago, -aginis. The etymology
is complicated by the initial component, which may indicate the influence of or assimilation to Latin
carrtts,'wagod.32 Nevertheless, Ammianus reports a Gothic term for a Gothic practice, and analogous
nominal compounds are attested in Germanic military vocabulary, albeit at a later date: Old English
bordhaga'shield-hedge', wihaga'battle-hedge'.33 There can be little doubt that this Germanic terminus
technicus entered Latin through contact with Gothic peoples in the Balkans. Fourth-century Roman
authors apply carrago only to the wagon-laagers of the Gothq a practice documented among various
Gothic groups from the mid third to late fifth centuries.3a Vegetius considered it a custom common to
'all barbariansi and similar practices are documented among contemporary Alans and Huns.35 In the
sixth century, comparable measures by Franks and Slavs are reported, though in each case probably
owing more to expediency than custom.36 Up until the end of the fifth century, both the term carrago
and the practice of using such wagon-laagers are attested only in connection with barbarian forces.
The Roman army's adoption of this procedure is first traceable in accounts of military operations in the
Balkans during the reign of Anastasius I (491-518), primarily against the Bulgars, even if the sources for
these events do not use the Gothic loanword.3T It is not until Maurice's Strategicore, however, that a Ro2e
30

31
32
33

tu
36
37

56

Lexica: Kowar.nwsrl 1844-9,1117782, str. touiba ='chevaux tress6s'; voN HAENTscH L939,153, 'tuh...Yakscheif, Fahne'; Crausor
7972, 464, s:t. tu:{,'the traditional Turkish standard was a horse's tail or a bunch of horsehair'; Houv 7984, 473, s.v. tzf: '(mil.)
Horse-tail crest/plume (sign of rank)'. Studies: Mnucrs 1951., 113; Poucne 7956,137-139; MarNcnsN-Hrlrnrt 1973, 410-11.
To my knowledge this proposition is seriously considered only by McCrindle 7897,360 n.2, 'Tupha is the Turkish name of
the horse-tail standard', thus WIxsreor 1909,349 (319.11): 'Gfigh), the Turkish name for the horse-tail standard, is probably
the word here represented by to0Qa'; whence this etymology is cited by Wolska-Conus 1968-73,111 323 n. 3; and turn by

Dubuisson-Schamp 2006, 1.2 16 n. 55.


Amm. Marc.31..7.7: ad carraginem, quam ita ipsi [sc. Gothi] appellant.

Krupr 1901, 347,365,387; Mlrrirscu 1968-9,1498;1978-9,111380; GnsrN 1998, 185 (somewhat confusing).
Boswonru-:Iou,nx 116, s tt. bordhaga; 7221, s.v. wtghaga.
Gothic carragoi Amm. Marc. 31,.7.5-7, 8.7, 72.17, 15.5; SHA GaIl.73.9; Claud. 6.6,8.2,5; Aur. 11.6. The practice is also alluded
to but not named in Claud. In Ruf. 2.124-9; De cons. Stilich. 7.94; Malchus frag. 18.1; Zos. 4.25.3; Marc. Comes s.a. 481. The
information in Zos. 1..45.1., possibly derived from Dexippug extends the evidence for such wagon-laagers back to the mid
third century. See generally Seofr 1938; NnpiioxrN 2002,9-15.
V"g. Epit.3.10.76 omnesbarbari. Alans: Claud. Bell. Goth.605; Amm. Marc. 31.2.18. Huns: |ord. Get.40 QlO). Cf. also Urbicius,
Epitedeuma 14 for unspecified paqpagou probably Bulgars.
Franks: Agath. Hisf. 2.4.4-5. Slavs: Theoph. Sim. Hist.7.2.4-9.
The first explicit evidence is j. Ant. frag.242.7 (Mariev : 311 Roberto = 214e7 Miiller), describing the encampment of the
large army of Hypatilus, magister militum per Thracias, at Acris near Odessus in 514 while campaigning against the rebel Vitalian and his Hunnic and Bulgar allies: oT qatorreberierat, to irc triv apalan yotqarc<':pa nqopaA6pevoq, 'he set up camp,
throwing around himself a palisade of wagons'. Cf. similarly Procop. BG 2.5.2-3,7.2, refering to Roman forces in Italy in
538. Somewhat earlier, it seems likely that the 520 'wagons ladened with weaponry required for combat' (plaustris armis ad
proeliandum necessariis oneratis) which accompanied the 15,000-strong army of Illyricum against the Bulgars in Thrace in
499 also functioned as a caffago, and likewise the significant but unspecified number of 'wagons ladened with weaponry
and supplies' (plaustraque armis atque commeatibus onerata) operating with the same army, 1Q000 strong, against the Gepids
in 505, cf. Marc. Com. s.aa. 499,505.
See

THE PHILOLOGICAL EVIDENCE FOR GERMANI IN THE LATE ROMAN ARMY

man author aPplies this term to a Roman military wagon-train. The Greek form rcaqny6q occurs first
in book XII in the context of infantry forces, where it refers to a train of heavier vehicles, distinct from
the army's main baggage-train (to0Aboq) and including wagon-mounted ballistae,which doubled as a
laager around an encampment and could similarly protect the rear of the infantry line in battle.3s It is
safe to assume the prior usage of this term within the sermo castrensis of the East Roman army,
dating
back at least to the early decades of the sixth century, if not before.
armilausa, armilausia,

d.

pp e

Aauo Lov

First attested in the later fourth century, Late Latin armilausa denotes a garment associated with soldiers.3e The earliest firmly dated instance occurs in the correspondence of Paulinus
of Nola, who in two
letters to Sulpicius Severus in 398-9 sharply criticizes the dress and character of a courier named
Marracinus employed by Severus. Paulinus finds fault with this wretcfu'who indeed initially astonished
us with his boot (caliga) and attire most unlike a monk's, while he is, moreover, no less red in the face
than his armilausa', subsequently declaring, 'so let him keep to himself his armilausa and his boots (caligae) . . -', apparently an evocation of military garb inappropriate to the wearert vocation
and status.ao
Other sources also attest to such coloured garments. A scholion on
fifth
satire
elucidates
the
Juvenal's
poet's phrase airidem thoraca,'a green corseleti with the gloss armilausiam prasinam.al The term
occurs
in two probably sixth-century papyri,both lists of clothing, which refer to a red and a green example;
neither document exhibits any obvious military connection.a2 Clearer definition comes towards the
end of the sixth century. An inventory of ttandard-issue' infantry equipment in Maurice,s Strategicon
(Iate 590s) specifies that all infantrymery 'should wear either Gothic-style belted tunics
or armilausia
(agpeAariora), short and as far as their knees'.a3 Around the same time, Isidore of Seville provides
the
only explicit description inhis Etymologiae, albeitmarred by spurious etymological deduction: talled
armLlausa in common parlance, since it is slit and open at the front and rear, and
closed only at the arms,
as if "armiclausa", the letter C having been omitted,.aa
At a much later date, in the ninth century, Photius compiled a series of historical notes and terminological glosses in which he remarks of another garment that it is,
Just like an a rmeslausion(riqpr;Aarioov)

38

3e
a0

Cf' Maur. Strat.

with commentary in RaNcn forthcoming. Distinction from main baggag e ttain: htrat.12.8.22.99:
Aotnov T o0Abov, 'the karagos and the ."Jt of th" touldos', cf. ato rz.E."ra..-t 2-12. Laager around an
encampment: 72.8.22-3, with diagram at 12.C. Protection of the rear of the infantry line: 12.8.18. For
the later Byzantine
orthography of raqgay6q, -6v, see LBG 764.
Du CaNcn 7678,1394;TLLrl674.6s-76;Mrni.escu 196g-9,14g4-s;Knarr.rnn 1996,113.,ls.
1,2.8.7.10-11

tov piv KaQaYov

rcai. 'rov

Paul' Nol4 ep'

e PL

17.1

armilausa ruberet; ep.22.7

67.2351t): quem quidem primo minime monachali caliga et zteste mirati, cum praeterea
facie non minus quam
PL 61.2548): sibi ergo ille habeat armilausqm suar; et suas caligas et suas buccas;with further remarks

on monastic and secular dress at 17'2. For the religious and literary context see Lrrrvuano -1977,73-5,whoclarifies
that Marracinus was not himself a monk but nevertheless became a foil for Paulinus' criticism of sartorial immodesty;
also

Tnour

"1999,129-30.

al Schol. adJuv. 5at.5.143 (Wessner).


a2 P' Milnch' 142.3 (Yl): eQF[e/a]Aauoov
1996,114-75.

Qodotov;

P.

Mich.684.17 (vlYI): agpagauoLv ngaowov (sic),

with remarks by Kneunn

a3 Maur'

Strat. 12.8.7.2-3: Ehe (c.rotriqrn lotOtrca ehe aqpeAauota ilouot rcovba, pr6lqr tci.rv yovatr^;v autr^rv
bei eoqerv
7-2.46-9 for (<ootaqtov as a generic term for any tunic worn with a belt. The translation
of DrNNrs 19g4,
138 wrongly attaches the phrase toming down to their knees' to the preceding Gothic
tunics rather than the armilausia.
The tenth-century Ambrosian paraphrase clarifies that the tunics, in contrast io the armilausia,'extenddown
to the feet,
(pi1qt nobriv bqrcovta), see LroNI 2003, 345. Dennis' definition of armilausia as 'short ones
[tunics] split up the sides, is also
misleading: on the contrary, Isidore of Seville explains that an aimilaus(i)a was split to the irorrt urrj
r"u. (see below).
Isid' Etym' 79.22.28: armilausa [var. lect.: armelausi, armelosaf oulgo oocato, quod anie et retro dioisa
atque

autofg' Cf' Strat.

aa

tum clausa, quasi armiclausa, C littera ablata. See Sorsn lgg},74_b.

aperta

est,

in armos tan-

57

PHILIPRANCE

[worn] on top of atmour'.45 Other Byzantine sources add little to our understanding. TWo different
versions of an apparently Middle Byzantine Passio of St. Boniface the Roman report that the martyr,
a steward in an aristocratic household, wore a Qoriorov aqpeAadoov or simply an aqrpraAariow (syncopated -Lov).6 Constantine VII Porphyrogenitus (r. 913-959) mentions the garment twice: first, as the
distinctive blue and white appar el of kombinographoi, a class of scribe; second, and somewhat incongruously, at the end of a list of poliorcetic equipment to be conveyed by a naval expedition.aT The exiguous western medieval sources suggest that the term came to denote a wider range of over-garments,
sometimes of luxury materials, worn by kings and monastic personnel.48
A late Roman armilaus(i)a or aqpeAaforov, attested from the end of the fourth century, therefore appears to have been a type of surcoat, typically but not exclusively worn by soldiers. Although Maurice
specifies a knee-length garment, the possibility of different lengths and designs cannot be excluded;
indeed the need for such specification arguably increases the likelihood of variation. Frequent references to conspicuous colour or colour combinations suggests that its function entailed recognition
or identity, which in military contexts may point to a degree of regularisation in military dress or
regimental uniformity.ae Philological scholarship has long identified a Germanic etymon. Mostly recently, Kramer has posited feminine substantive *armil(o)-laus1 or "arm(j)alaus6, an 'armless/sleeveless
(garment)'.50 The most obvious context for the passage of this Germanic loan into Latin is the Roman
adoption of a distinctive Germanic style of tunic, which would thus be another facet of the rich and
multifarious Roman-barbarian exchange in the sphere of clothing, decoration and fashion. Whether
this term in Latin and/or Greek retained its original specificity must remain uncertain. It is notable
that no Roman source indicates that such garments were 'sleevelessi but this may simply reflect the
fact that absence of sleeves was so intrinsic a characteristic of an armilausOa that such clarification was
redundant (just as in English 'shorts' are by definition 'legless'). Perhaps the most intriguing aspect
of the etymology is the existence of a Germanic tribe called the Armalausi or Armilausini, sparsely
documented in late Roman geographical opuscula and itinera, and located between the Alemanni and
Photius, Amphilochia, Qu.323 (Laourdas/Westerink VI.1 733.27-9):... droneq aqprlAanioov tov inav,,o rru.y 6nLan. Photius'
principal source in this section was John Lydus' De magistratibus; the reference to an aqpqAariorov appears to be his own.
A misplaced version of the same gloss appears in the late tenth-/eleventh-century Etymologicum Gudianum (De Stefani I
200.8): Agpdoriotqtov: naqa to dqpa to ini [cod. d1: enavc-r] t6v 6nArov. The term AqptAoriotqrrov (Latrn Armilustrium), a
festival of Mars, here confused with aqprlAarloLov, is found in the same section of John Lydus' work. Modern scholarship
exhibits some confusion with regard to the lexical literature. Du CaNcn 1678,1394 ascribes the gloss inavco trov 6nArov to
the Suda, inwhich it does not occur. Similarly, Knarvrnn 1996, 114 cites the Su da lor partly tlire same gloss: agptAariorov naqa
to dqrpra,ro inoLv,,o rCov 6nL.av, but does not provide a specific reference.
The first version of the anonymors Passio Bonifutii Romani was published from an unspecified Vatican manuscript by Brcor
1680,310-324 at 321: Qodotov aqrpeAaiorov. An undated Latin version of the same Passio was previously edited from seven
codices in Rome by Hor.srr 7663,27: coccinea armelausia. The relationship between the Latin and Greek versions has not been
established. The juxtaposed texts of both Bigot and Holste were published in Ruinart 1713,290, thence reprinted 1859, 331.
The second Greek version is preserved in codex Athos Iviron 4887.767, where it is ascribed to Ephrem the Syrian. The text
has recently been included in that author's oeuvre: Pnnarsrzoras 1998,187:198 at 196.13: aqpaAauow. Given that so much
of "Ephraem graecus" is pseudepigrapha of a much later date, the ascription can only be doubted.
Const. Porph. De cerim.1.80 (71) (Vogt II 153.19): Qoqrorivt<ov aut6v tiov roprpLvoyqaQrov aqpe),ai;ota pdvetov rcai,Aeur6v;
2.45 (Reiske ll 67O.78): aqpeAafoLa : Har-noN 2000 225.727 with remarks at 276.
Liber Pontificalis (V. Greg.Il) 97.22 (Duchesne | 408b.7): . .. ut lrex) quae fuerat indutus, exueret ... mantum, armilausiam, balteum,
spatam atque ensem deauratos (in reference to Luitprand, King of the Lombards 712144). A charter (a. 849) of Charles II the
Bald to the Abbey of St. Martiry Totl-s: Dedisse se fratibus S. Martini Antoniacum aillam ad eorum armelaasas (codices: armelanias CDG: corr. armelausas E), see Glnv et a\.1943-55, I 300-303, Ns 113 (302.7-8). The broadening of meaning is also reflected in
glosses: CGL V 338.51: armilausia sercae (Knanrn 1996, 115 restores ser<i>cae, surely correctly); 589.44:, armilaus<a> scapulare
monachorum. Older studies and lexica cite a charter (a. 605) of ,4thelberht, King of Kent, which gifts to the monastery of SS.
Peter and Pauf Canterbury, luxury items that Pope Gregory I had sent to him for this purpose. The list includes armilcaisia
oloserica, accusative and so seemingly neuter plural, apparently silk vestments. This document is demonstrably a later forgery, however, probably compiled in the second half of the eleventh century; see KrI-rv 1995, 13-1& Ns 3, with discussion of

dating at lxiv-Ixv.
Maurice does not specify the colour ol armilausia, brt at Strat. 12.8.4.2 he requires all the shields in a regiment to be of

uniform colour.
Knaunn 1996,

58

114

n. 10 summarises the possibilities, with older bibliography.

THE PHILOLOGICAL EVIDENCE FOR GERMANI IN THE LATE ROMAN ARMY

Marcomanni in the region of the Bohemian Forest and Fichtel Mountains. I have nothing to add to this

particular puzzle.sl

CATEGORY

2:

NOMINAL BORROWINGS

framea

The word framea is first attested in Tacitus' Germania, written c.9g which provides the only explicit
definition: 'The [Germani] carry speart or frameae to use their own term, with a narrow and short
iron head, but so sharp and easy to use that with the same weapory as circumstance requireg they
may fight either at close-quarters or at a distance'. Tacitus draws a contrast with'larger lances' (maiores
lanceae) used by a select few. Having established the meaning of framea - apparently a shorter, multiPurpose thrusting-throwing spear, employed by both infantry and cavalry - he proceeds to use only
this term throughout the rest of his work whenever he refers to the spears of the Germani.52 It is not
entirely clear whether Tacitus introduces a hitherto unknown term reported to him by his source(s) on
Germania or employs a word already current in Latin and familiar to Roman readers through previous contacts with Germanic peoples. In any case, framea soon afterwards crops up in Juvenalb verse,
and slightly later appears in the antiquarian compilation of Aulus Gellius (late second century) and the
juristic writings of Ulpian (early third century), both of whom are concerned to compile and/or define
the terminology of projectiles. From these instances it seems that framea became a generic Latin term
for a spear or javelin.s3 A few later authors chose to apply this word to contemporary spearE with a
doubtful or uncertain degree of specificity.sa From the later fourth century onwards, however, framea
occurs with far greater frequency in diverse theological writings, particularly biblical commentaries,
in which it clearly has no technical meaning and even its basic identification becomes subject to widespread confusion, to the extent that many patristic writers believe a framea to be a type of sword.ss
Despite Tacitus' unequivocal statement that framea is a Germanic word Qpsorum oocabulo), its etymon remains obscure. Correspondences are commonly adduced with a Germanic causative verb*framjan, 'to bring forward, advance, implement, promote, further, accomplish' (> Old Saxon
fremmian; Old
Englishfremman; OHG fremmen; Old Norse/remjn); and adjective *frama-, with the basic sense of 'prominenf pre-eminent, protruding, projecting' (> Gothic adv. fram, 'further'; Old Saxon adv. fram, 'further,
fartherl Old English ad| fram, 'pre-eminent,bold'; Old Norse adj. framr, 'prominenf protrudingi adv.
fram,'forward'; Old High German adv. fram,'further'). Scholarship has long drawn a morphological
parallel with a (possibly East Germanic - Burgundian?) runic inscription on a probably third-century
spearhead found at Dahmsdorf, Brandenburg, which rcads ran(n)ja, 'hitter, strikeri one of several such
inscriptions, widely distributed in Germanic territory, which point to a common Germanic custom of
naming sPears with a personalised and possibly amuletic nomen agentis. Accordingly, the most plausible etymology of framea posits an analogous Germanic masculine agentive noun"framja (older studies prefer masculine "framjQ with a sense akin to 'thrusteri but in this case necessarily a generic
designation for a spear or a type thereof. In Latin the loan framea would have acquired a feminine
gender by analogy to existing Latin terms in the same semantic sphere (hasta, lancea). The dearth of
correspondents in Germanic languages or cognate elements in Germanic personal names implies that,

s]
s2

s3

'n
ss

in TLLll614.77-80; Scu<iNrelo 1911,,28-9, ssr. *Armilausi.


Tac. Germ. 6.1: Hastas oel ipsorum oocabulo frameas gerunt angusto et breai ferro, sed ita acri et ad usum habili, ut eodem telo, prout
ratio poscit ztel comminus z.tel eminus pugnent. Cf. Germ.11,13,74,18, 24. Commentaries: Mucn 1967, 135-9; LuNn 1988; Pnnr 1990,
148; Rrvrs 1999, 738-9. For the possible design of a/ramea see Frrsrcon 191Q citing older literature; Muca 7967,136-9; Srrurn
7995.
See references

Iuv. S at. 13.79; Gell. NA 10.25.2; Dig. 43.16.352 (Ulpian).


E.g.Eucherius, Lnstr.2147.72. (CCSL 66 194.153-5); Greg. Tur. Hist, Franc.3.15;2.46.
See references assembled in TLL VI.1 7239.76-1240.53. Corresponding confusion emerges in the late lexical evidence for
other weapons; e.g. Hesychius Lex. (153: (fuvrl '6Aoo(b1qrov ar6vtrov t A6yX1 11 ona01 11 palarqa (Latte II 261).

59

PHILIPRANCE

if Tacitus does report

a lexical novelty, his informants had chanced upon an uncommon term, at least
relative to the numerous other Germanic words for spears.s6 There is no evidence to suggest that the
initial or increasing incidence of framea in Latin texts corresponds to any actual transfer or imitation of
Germanic weaponry nor thatframea was ever a component of Roman sermo castrensls. The absence of
correspondents to framea in Romance suggests that it did not penetrate Vulgar Latin.sT Rather it seems
to be a question of a purely linguistic loan, initially (and unusually) motivated by the erudite lexical
interest of literary writers, which merely supplied yet another useful synonym to the existing Latin
thesaurus of words for'spear'. One may legitimately suspect that Tacitus' ethnographic excursus on
the Germani was the main (and possibly sole) conduit for the entry of this loanword. Its wider currency in later literature should be partly attributed to its presence in the works of canonical authors
like Tacitus and |uvenal.

fulcum, Qoohovss
Maurice's Strategicon applies the term QoOArcov to a multi-tiered shield-wall constructed by the front
two or three ranks of a close-order infantry formation, depending on tactical circumstances.se The text
preserves the wording of the contemporary Latin command ad fulco,'into a fulcum', which indicates
that QoOArcov is a Greek transliterationof Latinfulcum, aterm not otherwise directly attested in Latin.60
This term appears to have enjoyed an institutionalised currency as part of standard late Roman military vocabulary; at no point does Maurice deem it necessary to explain or contextualisefulcum in a way
that might signal its foreign origin or anticipated unfamiliarity to the reader. It follows that there is no
reason to assume a particularly eastern or'Byzantine' scenario for the origin of this loanword purely
on the basis of its initial appearance, transliterated into Greek, in a late sixth-century East Roman treatise; rather the evidence points to wider chronological and geographical horizons. In additiory while
derivative Byzantine military literature reproduces Maurice's description and terminological usage of
Qo0Arcov,61 most later texts employ this term in a significantly different sense. Tenth-century tactical
manuals apply QonArcov to a close-order formation, typically of cavahy, deployed in support of raiding or foraging parties.62 The two instances of Qo0Arov in historical texts, Theophanes' Chronographia
(completed c.810-14) and'Theophanes Continuatus' (c.970), refer more broadly to generic'divisions' or
'formations' comprising a battle line.63 These variant meanings, although attested only at a later date,
allow for the possibility thatfulcumlQo0Arcov always had a broader semantic range than Mauricet spe-

56
57
58

59

60

61

62
63

60

For detailed discussion see Knausn 1936; Knaur 1965; Mucu 1967, "135-6; Kuur.r 1972, 44; Mrrwnxn 1995; GnneN 1998, 785-6.
Bniicu 1913,19.
I have discussed this term at length in RaNca2O}4b; a summary of the evidence and arguments is offered here, with some
supplements and ref inements.
Maur. Strat.72.A.49-60,8.74.9,16.30-38,24.13 with commentary in Raucn forthcoming.
Marrr. 9trat.12.8.76.30. This command is variously corrupted in the codices: AaQouArco MLp: a6 QouArcr.r VN: abeQouArr.;
P. The tenth-century Ambrosian paraphrase renders the command into contemporary Greek:
Qo0Arcov rroqoate. Scnrrer:x 1664,526 restored A6 Ooriz\rc<lp. Drrqxrs 1981, 442 app. crit. reconstructs ab-QouArov, but the Greek termination -ov
is certainly invalid in a Latin command. Initial AA is secure and clearly AA in MLp is due to misreading in a majuscule
ancestor(s).WhereallcodicessuggestLatinfulco,however,
adfulcumcannotbeassumed.Thereading/zlcoreceivesfurther
support by the corrupted ami fulco in the recapitulation of commands at 12.B.24.13. Furthermore, within the same treatise
there are other examples of Latin commands, common to all codiceg in which the preposition ad is followed by a noun
ending dative -o where accusative -urz would be expected: e. g. ad conto clina (12.8.16.78) and ad scuto clina (ibid. 80). Mrn,i,rscu
7969,263-5 sttggests that this divergence from classical forms reflects the loss or confusion of case endings in spoken usage,
at least in the context of military commands; thus also MrHi.escu 1970,330.
Leo, Problemata 72.29; Tactica7.48, 54. See RaNca 2004b, 313:27.
See evidence and discussion in RaNcr 2004b,321-5 with bibliography.
Theoph. Chron. 378.15-16 (De Boor): naq..tafaro tqtoi. QodAxoLq , 'arrayed his forces in three Qoio/tra', in reference to the
Persian commander Rhazates at the battle of Nineveh in 627. Theoph. Cont. 6.41 (Bekker 460.2-7): rag naparc.lug xai
6uvapetg rcaiTa QoOArca to0 a06ou Xappaba,'the formations and forces and QoOArca of the godless Hamdun', namely
Ali ibn-Hamdun in the late 950s. See remarks in RaNcn 2004b,310'13.

THE PHILOLOGICAL EVIDENCE FOR GERMANI IN THE LATE ROMAN ARMY

cific usage implies, a prospect enhanced by the evidence for cognate terms in Germanic languages (below) and, more generally,by the variability of other items of Latin military jargon (e.g. cuneus, drungus).
Although several etymologies have been proposed, only a Germanic derivation stands up to scrutiny. In his annotations to the editio princeps (1664), Scheffer first drew attention to the resemblance
betweenfulcum and modern German Volk, an association accepted by Du Cange.6a The extreme rarity of Scheffer's edition, the only published text of the Strategiconbefore the later twentieth century,
meant that fulcumlQotArcov was overlooked in all previous surveys of Germanic loanwords in Latin.
The etymon is identifiable as Germanic neuter noun*fulka-. Correspondents are attested in every Germanic language except Gothic and convey different but overlapping political-military-legal spheres of
meaning: Old High Germanfolc, 'people, crowd, army, hos! warband'; Old Saxon/olk,'troop,divisionl
Old Frisian/olk,OldEnglishfolc, 'people, host'(cf. derivative Old English gefylce, 'troop, division,); Old
Notsefolk,f6lk,'people, hosf troop', with the derivative nounfylking, 'battle array,host', andverbfylkja,
'to deploy, to array in formation'.6s The wider influence of this Germanic root is evidenced by loans in
Lithuanian: pulkas,'bartd', and Old Church Slavonic: rrz'r,Kt, (< Proto-Slavic*p.olko), 'warbandi which
later evolved into the standard term for 'regiment' or'military unit' in modern Slavic languages.66
Although/olklfolc acquired multiple semantic functions, the earliest evidence suggests that inlilitary
contexts this term less often denoted an 'army' (a sense supplied by her- forms) than its constituent
sub-divisions - contingents, warbands or combat formations - when deployed for battle. This view is
consistent with the testimony of Old Saxon and Old Norse, the two most conservative Germanic lanSuages. Old High German glosses equate folc to Latin cohors, manipulus and. cuneus, but never exercitus.
Similarly, Old English gefylce occurs as a gloss for Latin manipulus, cateroa and legio.67 The absence of
a correspondent in Gothic may point to a West Germanic linguistic sphere. It is tempting to see the
fourth or early fifth centuries as the most likely period for the loanfolk>
fulcum,when numerous units
of auxilia were recruited from peoples living east of the Rhine. It seems probable tha! writing in the
590s, Maurice used a word that had gained popular currency in the Latin-based argot of late Roman
army.
The precise dynamics of the loan are not easily clarified, not least because the linguistic evidence
points to different conceptions among and between Germanic- and Latin-speakers. In the Strategicon,
the earliest and unique Late Antique witnesE it is clear that Maurice applies the term
eooArcov specifically to the thield-wall' constructed by the front two or three ranks, and not to the compact close-order
formation as a whole.68 In accordance with late Roman tactical doctrine, the first and second ranks
were typically occupied by junior regimental officers, who wore additional protective equipment and
acted as file-leaders' (Aoyayo(, nqr,srooraran) in combat.6e As outlined above, however, Germanicfolkl
folc applies more generally to a battle array or a body of troops. The parallels with Maurice's eo6Arcov

66
67

ScnrrreR 1664,497-8, 'Est autem QoOArot a Folck, quod turbam condensatam denotat, cumprimis militum,. Accepted by Du
CaNcr 1688, 7694-5, s.v. Qo0Axov See also MrH.i.Bscu 7970, g0Z C Strat. 12.A.7).
Poronuv 1959,799-800, s:t. pel- lDlexplains the underlying Indo-European motivation. The Germanic evidence is presented
by Hrnoro 7941., 184-9; Klucs-Srssolo 1989,768 s:t. Volh ErrnrsuaNN fSeS,
73-8; Gnrrr.r 7998,9O-95;See also BoswonrgTottax296, str. folc,394, s:r. gefylce; CLrl.snv 1957,767, str. f6lk (ll);779, s.vv."sp.
fylking, fylkja.Mod,etnErench foule, trowd,, has
the same derivation via OldEtenchfolc, foulc.
E.g.Polishpulk, Russian and Bulgarian noax, Serbian ny& Czech and Slovak pluk.
Hnnoro 1947, 184-9; Gnnnu 1998, 90-92. Heroid and Green perhaps underestimate the currency of the meaning,army,. The
present writer finds the evidence adduced in the Hildebrandslled to be inconclusive. They allow a single testimony in tire
Old
High German version of Isidore of Seville's De fide 4.11: celestis exercitus > himilisca
foL. Ho*"r"r, tley overlook instances,
not registered in Boswonru-llort-sv296 s.v. folc, in which/olc renders exercitus in the Old English Orosius, see Brrrrv 19ge
e.9.26.12, 33.17, 45.24, 46.22.
MaJJt. Strat. 12.A.7.52-3: Kai.T ov ptiv

nqritov rcal beu,reqrov rai- tq(tov Lxaorqearclaq eiq eooArov rca}Lorao1at, rout6otw
iQ' tv orcoutaqLv, And the first, second and third man in each file are to form themselves into afulcum,that is to say one
shield upon another'. Similarly, at12.8.76.30-3$ the order ad
fulco reqluires only that the men of the first two ranks link their
shields together; the whole formation has previously closed up its rinks a.rd hles upon the order iunge (ibid.20-29) and
the
order to form a fulcum doesnot require any further action from the rest of un it; cf.12.8.74.8-9;24;,2-13ior the same sequence
Ev

of orders and manoeuvres.

Maur. Strat.12.8.4.5-8, 16.37-2- 54-5, cf. Agath. Hlsf. 2.8.5 with RaNcr 2004b, 273-4. See also
JaNNrenr 2004a.

67

PHILIP RANCE

are therefore, at best, inexact. In contrast, the alternative uses of QoOArcov documented in Middle Byzantine sources exhibit closer correspondence to folklfolc, which allows for the possibility that the usage
found in the Strategicon reflects merely one of several meanings of fulcumlQotArcov current in the late
sixth century. Furthermore, the evidence for a long-standing Roman tactical tradition of close-order
formations fronted by a 'shield-wall' or thieldJinkage' (testudo, yeLtisvtl, ouvaonrop6q) suggests that

the motivation of the loan folk > fulcum did not involve late Roman imitation of a Germanic style of
fighting but rather a nominal borrowing for a shared aspect of military culture.To This in turn raises
further complications. First, shield-walls are attested in the tactics of some Germanic peoples of the
medieval period, notably Anglo-Saxons, in the ninth and tenth centuries, and Scandinavian armies, at
a similar historical horizory though for the most part reported in much later saga literature.Tl There is,
however, very little evidence that ancient Germani formed shield-walls of this type in battle, certainly
to a degree that could be considered regular or characteristic of their style of combat.72 More explicit
documentation of Germanic shield-walls emerges only from the mid sixth century, a period in which
some scholars have discerned a general shift towards close-order infantry tactics in western Germanic
armies.73 Second, even when shield-walls are attested much later in the history of Germanic peoples,
folklfolc and its cognates are never employed to describe this practice. If Late Latinfulcum was borrowed from Germanic peoples to describe a Roman'shield-wall', how do we explain their use of very
different terms to denote the same phenomenon: Old Norse skjaldborg ('shield-fort'); Old English scildburh (shield-fort'), scildhre6da ('shield-cover'), scildweall ('shield-wall'), bordzneall ('shield-wall'), bordhaga
(thield-hedge') or wihaga ('battle-hedge')?7a Even accepting that some of these terms may reflect poetic
usages or metaphorical expressions, we have already seen that, when necessary or appropriate, the
Romans did adopt Germanic nominal compounds of this type as termini technici in their own military
vocabulary: Gothic *carr(h)dgo ('wagon-hedge') > Latin carragoIf definitive answers remain elusive, the evidence permits some tentative conclusions. On the one
hand" the scarcity of Germanic loanwords in Latin military vocabulary suggests that, in this rare instance, Germani recruited into the Roman army continued to employ a word from their own language(s)
precisely because it retained a meaning and significance for them in their new cultural setting. That
is to say, they applied the Germanic word folklfolc to the Roman deployment that most resembled
their own way of fighting - a close-order infantry array, fronted by better-equipped and more-experienced'professional' warriors. The underlying linguistic motivation would therefore be recognition
and familiarity, not innovation. On the other hand, to judge from Maurice's usage, at least some Latinspeakers inferred that this new Germanism current among auxilia had a narrower or technical meaning associated with this particular deployment and came to understand the'shield-wal1' as intrinsic
70
71

See Rer.rcn 2004b, 295-304; also Wneer-rn 2004 350-53, contraSpnrost 2004a, 103-8.
See Harsalr 2001., 177-9, 783, 194-7,208-10. Attempts to reconstruct the operation

of a generic 'shield-wall' in post-Roman


Germanic warfare can be found in UNornwooo 1999, 89-91,,129-131; SrBpnnNsoN 2007,36-40.
72 I know of only two classical loci which possibly refer to ancient Germani employing a'shield-wall'. First, Caesar's account
(BG 1,.52.4-5) of his victory over the forces of Ariovistus in 58 BC implies that their'customary phalanx' (ex consuetudine
sua phalange facta) was protected by a barrier of shields; this at least seems to be how this passage was interpreted in
much later derivative texts (cf. Cass. Dio 38.49; Oros. 6.7.8-9). Second, Plut. Mar 26.4 rcpofis that at the battle of Vercellae
in 101 BC the Cimbri, when discomforted by the great heat and dust, 'were holding their shields before their faces' (todq
0ugeouq nqopaAA6pevot nqoT riv nqoo<inov). Plutarch (or his source) may have misunderstood the motivation for this
conduct; he earlier alludes to'linking shields' as a potential tactic of the Teutones in a battle the previous year (bid.20.6:
tov ouvaonrop6v). One could adduce equally meagre comparative evidence relating to Gauls: Livy 7O.29.6-7 (Cisalpine
Senones in 295 BC) and implied by Caesar, BG 1,.24.5,25.2-3 (Helvetii in 58 BC).
" E.8.Agath. Hlsf. 1.21.6-8;2.8.8 (Franks-Alemanni); possibly Procop. BG 1..29.35-6 (Ostrogoths), with RaNcr 2004b, 290-95;
jexxIano 2004b,1077. See remarks of Harsar.r 2001,,208-10 on general trends in western Germanic warfare c.600. A representation of a two-tiered shield-wall is one possible explanation of the depiction of warriors on the lyre discovered in an
Alemannic elite grave at Trossinger, dated by dendrochronology to c.580; see TuruNr-Gnossxorr 2006.

7a Crrasrv

1957, 551', s.v. skjald-;Boswontu-rTor.r-rn 1L6-77, sstv. bordhaga; bordweall, 831,, strv scildburh, scildhreida, scildweall, 1221,,

s.v.zttighaga.SomecontemporaryLatinaccountsof Anglo-Saxonand/orVikingarmiesincombatreferto testudo,-ines,wlnict.


probably signifies a generic close-order battle formation, although this classicizing term may have been selected for its
implication of a 'shield-wall', e.g. Asser, De rebus gestis Aelfredi 37-8,56.

THE PHILOLOGICAL EVIDENCE FOR GERMANI IN THE LATE ROMAN ARMY

to the meaning of folk > fulcum. There are too many imponderable factors to determine whether the
'institutionalisation' of the Germanic-derived termfulcum necessarily reflects an overwhelming presence of Germanic warriors in late Roman armies, though this linguistic development is consistent with
the view that ethnically Germanic infantry units were favoured for and more frequently employed in
this style of fighting, just as Roman military recruitment had always targeted different ethnic-cultural
groups according to their combat specialisms.

CATEGORY

3:

UNCERTAIN WHETHER A LOAN-OBJECT OR A NOMINAL


BORROWING

bandum,

pav\ov

During the first half of the sixth century, Late Latin bandum, most frequently documented in Greek
transliteration pav6ov, emerged as the most common term for a military flag or standard, replacing
aexillum and signum in the Latin-based argot of the East Roman army. This development is first reported in contemporary Greek historical sources, some of which identify pdvbov as a Latinism.Ts The
alternative contention of Kramer that the term first evolved in Greek, and was only later Latinised as
bandum, is demonstrably misconceived.T5 The term is later also found transliterated into Syriac.77 An
East Germanic derivation has long been recognised, certainly the Gothic feminine nowbanilwl,'sigrt,
markeri presumably a consequence of increased Roman contact with Gothic groups, as enemies, allies
and foederatl, over the course of the fifth century.Ts Variant bandus is a later development peculiar to
See Cosmas Ind. 11.5; Malalas, Chron.L&.S9 (Thurn 387.78),60 (388.34 389.40),65 (391.35); Procop. 8V2.2.1;Theoph. Sim. Hisf.
3.4.4, 6.4;7.3.3; Chron. Pasch.707.78 (Dindorf); Paul. Diac. Hist. Lang.1.20; Theoph. Chron. 260.10, 319.5, 322.2 (De Boor). See
also later glosses: CGL 1128.40;Y 505.7; Glossario di Caoa 17; GlossarioVaticano 12; Lexicon in orationes Greg. Naz. s.v. <nivQ1pta
(Sajdak 186.4-s).
See Kramer 1987, esp.200-203,207;1996,115-16,126, reiterated 201''1., 4L-2; followed by Morlicchio 2002,3b8;2004, tO6-2. Citing Procop. BV 2.2.1: to or;peiov 6 b1 privbov rcaAo0oL Pa;gaioL, 'the standard which the Romans call ...', Kramer insists
that'PropaLot must be understood as 'Griechen' and thus Procopius here draws a distinction between classical and contemPorary Greek: or;pelov and pdvbov. Kramer opines that only if Procopius had employed wording such as tf Aa.rlvov
Qovr,1 could this term be considered Latin. A wider examination of Procopius' writings reveals thii interpretation to be

77
78

erroneous. On the contrary, leaving aside references to place names, in the many instances in which Procopius employs the
epexegetical phrase rcaAo0oL'Pr.-rpatot his purpose is to introduce or explain a Latinism, the use of which in claisicizing
Greek historiography required justification or circumlocution: e.g. BP 1.22.5: rivrov ydo ta ircatov rcaAoooL ?copalo; fo-r
the Romans call a hundred "centum"'; BV 1,.77.4: Qoi6eqa yag tag nqog touq rroAe p[ouq o'zovbag rcaAof,roL PopaioL, ,for the
Romans call treaties with their enemies "foedera"'; ibid. 6.10: MegrcoriqrLov ... yaq tov 'Eqptrlv raAouoL'propar-or, 'for the Romans call Hermes "Mercurius" l see thus rcaAouor ?t rpaiot at e.g. Bp 1.24.72 (quaestor);2.7.1,5 (a secretis),2L.2 (silentarius), 22.7
(referendarius); Bv 1.4.7, 11.6 (domesticus),70.3 (praetoi, 17.2;2.20.12-13 (optio);2.3.28 (December),2G.27 (casula); BG 1.22.10,22.17
(aioarium); Anec. 28.6 (tabellio); Aed.3.3.14 (d.uil; 4.6.16 (pon s). Furthermore, elsewhere procopius expressly contrasts
the Latin
language of ?opaiot with Greek terms or forms: e.g. BP 1.29.26: rcAeLooriqraq 6AAr;vi(ovteq ... LaAo0or P<opaior, ,which
the Romans call "kleisourai" when they translate into Greek (i.e. Latin clausurae)'; and even more explicitly: BV 7.27.2: ...
rcaAo0ot'Popa[ot, ou t1 oQttdq qylrboorl, aAAa . .. lAAr;vi(ovteq, 'the Romans call [this place] .. . notin their own tongue,
but using Greek'; cf. also BG 3.1.29. There is no reason to believe that Procopius' reference to pav6ov is likewise anything
other than a Latinism: bandum. In his subsequent explanation at BV 2.7O.4:6v 6r; pavboQ6qov rcaAoOor ?co;.raior, 'whom the
Romans call a bandum-bearer', his use of Greek -oQoqov rather than Latin -tQega (<bandifer), awkward to decline and entirely unprecedented in the genre, is unremarkable and does not controvert the preceding analysis. In addition, Theophylact
Simocatta similarly remarks atHist.3.4.4: ta crypeia ... ri t11 natqicp
Qov1l privba'Popaior rcatovopa(ouow, 'the standards
... which Romans termbanda in their ancestral language', thus also 3.6.4;7.3.3. Again, Theophylact's references to the
'ancestral' or'native' tongue of the ?copalot serve to signal a Latin word or phrase: e.g.2.4.7 (*tuldum);2.15.9 (,,torna, torna,,);
6.9.74 (sculca);7.2.6 (the language of orders, at this date still issued in Latin); see Barowrn 7977;Whitby 1982. More generally,
Kramer's contention that'Pr.rpraiot must indicate Greek-speakers is further undermined by the contrast of 'P<.rpatotl rcal
'EAAlvLoti, 'in Latin and Greek', in the Strategicon (1..8.3-4;12.8.7.4) and the author's decision
to use 'Prrpaircaiq A[ tos,Latin words' (pt.29-30).
E.g. John of Ephesus, HE 6.2 (twice).

Etymology and development: Ksupr 1901, 348-9, 368-9, 387; Hrnarus 1902, 264; Bniicr 1913, 58, 162; ScHnayu 1954,
MrHi.escu 7978-9,111371; Rssrsrl-r 1984,162-9; Kna^asn 1987;1996,115-16; Monr.rccnro 2002,35g.

250;

63

PHILIPRANCE

the Latinity of Ravennate and Lombard writers.Te One must entertain the possiblility, even likelihood,
that a Lombardic correspondent to Gothic bandwd later reinforced the currency of this term in ltaloRomance.so

The loan bqndwo > bandumlpalbov subsequently inspired the internal development of both Latin
bandifer and Greek pavboQ6qoq in the sense of a generic standard-bearer, whether it be a generalt personal insignia or the regimental colours.81 Soonbandum came also to apply metonymically to the body
of troops serving under the flag, a semantic development similar but not directly analogous to oexillum
> aexillatio. Currently this usage is documented with certainty only in connection to cavalry units and
is not attested in a firmly dated context before the Strntegicorz, in which pavbov designates the basic
administrative and operational unit or 'regiment' (synonymous with rot1yvo- and aqt0p6q = numerus).
Maurice thus employs the term pavbov in the sense of both'standard' and 'regimenti sometimes confusingly in close proximity. The extensionof bandum to designate a unit was therefore a purely linguistic process and did not entail organizational innovation.s2 The specificity of bandum in the sense
of 'standardi and thus the motivation for the loary is less easily discerned. As the earliest witnesses
seem to apply bandumlpavbov indiscriminately to all styles of Roman (and even Persian) military flags
or standards, it is difficult to determine whether the loan originally took place in the context of the
Roman adoption of a specific type of 'Germanic' standard or was merely a generic nominal borrowing.s3 On the other hand, the specifications for pavba inthe Strategicon appear to delineate a particular
design of 'flag', in the modern sense, rather than a aexillum-style'banner' suspended from a horizontal
cross-piece, as used by the Roman army in previous centuries.sa The slim evidence and non-technical
vocabulary for Gothic standards in late Roman sources offer no assistance in resolving this question.85

D
m

BnowN 1984,84n.8,89-91; Knaurn 1987,204-7.

Knaurn 1987,203-4;2071.,42; Monrrccnro 2002, esp.358. Later medieval instances otbandum and derivatives are collected
by Du Caxcn 1678,1547.

84
85

64

The term pavboQ6gog is first attested in Procop. BV 2.10.4:6 . .. to olgeiov toi otqatrlyoi ev taiq taqratd[,eow eko0dq
at DrNNrs 1981,
Qqew, ov 6q pav6oQ6qov raAo0or ?copraior; see the definition at Maur. Strat. 1..3.212. with other instances
520 index graec. s.v. pav6oQ6qoq. In the Latin harangue at Maur. Strat.3.5.7, the standard-bearer is addressed asbandifer
(pavbrQeq in M). The wider currency of this form is uncertain; it is unknown in the Latin terminology of thelater exercitus
Italiae, whict. simply Latinised pavboQ6goq asbandiforus, see Brown 1984, 59-60.
For pavbov in the Strategicofl see DsNr.rrs 7981, 520, index graec. Alternatively, Knaurn 1987, 202; 1996, 116 n. 20 wishes to
identify the earliest instance of pavbov with the sense of 'Heeresabteilung' in Malalas, Chron. 18.59 (Thurn 387.78), who
reports that the Persian general Azarethes peta lleqoLrrlS porl0eiaq, tlrov peO' lautoO rcai pavbov paoilLrc6v, invaded
Roman territory. However, given that every other instance of pavbov in Malalas' work, all occuring in the same book (18.60
1388.33, 389.401,65 [391.35]) clearly refers to a flag and not a unit, the meaning here should be'with a Persian force, taking
with him also the royal standard'; see thus the English translation in Jrrrnrvs et al. 1986,270. For bandum as 'regiment'
see the possibly sixth-century inscription from Odessus (Moesia II) in BrSrvrIrv 1964,60-61, Ns 89. Instances in the early
seventh century include P. Oxy 2010.1-2 (618); Miracula S. Anaslasii Persae 13-14 (Flusin I 143,145). For a possible instance of
bandum (='unit') prior to the Strategicon see the fragmentary inscription of 531 from Pylai published as AEI995.1427 =lK-32
136: SEG XXXVII 1023; XLII 1124 which Zuckerman restores as (ol,rgattoritlq 6euti[qo]u [p]riv6ou Kootav[t]Lv1ar6v),
apparently'a soldier of the secondbandum of the Constantiniaci'. See ZucKERMAN 1995,233-5.The restoration is problematic
for two reasons. First, all other sources, including the Strategicon, treatbandum as a term for the'regiment' itself (in this
case an auxiliumpalatinum), rather than an organizational sub-division of a regimenf as if'second battalion of ...'; indeed
Zuckerman himself notes that such a division 'en compagnies ... est fort remarquable Q35)'. Second, this would also be a
unique instance (where the troop-type is specified) olbandum applied to infantry.
Explanations and glosses imply broad equivalence with other terms: CGL ll28.40; V 505.2 bandum o[yvov (MS Atyvov);
Procop. BV2.2.L: to orlpeiov 6 6r; pavDov rcaAo0oL?<.rpatov ibid.70.4:6...to or;peiov ... eiroOdq Qdgew, 6v br1 pavboQ6gov
rcaAo0oL'Pr^rpaior (Procopius refers in both passages lo the oexillum pruetorium, the commander's standard); Theoph. Sim.
Hist.3.4.4: ta orlpreia ,ro0 lleqorrco0, & tfl natgi<p Qrovr,l pavba 'Parpaiot raT ovopa(ouow; Paul. Diac. Hist. Lang. 1..20: oexillum quod bandum appellant lLangobardil; Lexicon in orationes Greg. Naz. s.v. oriv0r;pa (Sajdak 786.4-5): orivOqpa to orlpeiov
rcai oJppoAov toO o,rqa,roO, 6 raAoOoL pdvbov' ta rcaAoripeva naqa 'Pop alotc, rttyva rcai privta (sic) (a variant of the same
gloss is cited by Du CaNcr 1678,1547 as Schol. ad Greg. Naz. Or. 4.66, from a different and seemingly still unpublished
collection in 'Cod. Reg. 3183', now Parisinus gr.2314).
Byzantine flags: DrNnrs 1982,57-9. The form of earlier Romanoexilla: voN Douesznwsru 1885 76-80.
E.g. Amm. Marc. 31.5.8 (oexilla); Sid. Apoll. Carm. 2.232-3 (fuacones); Ennodius, Paneg. Theod.7.32 (signa); Procop. BG 3.24.23
(orlpeiov).

THE PHILOLOGICAL EVIDENCE FOR GERMANI IN THE LATE ROMAN ARMY

punga, ftouyyiov and oeA\onobyytov

In the

Strategicon Maurice uses the words nouy"ylov and oeAAonofyyLov (also syncopated
oeAAonoilyyLv) to denote the standard-issue saddlebag of an East Roman cavalryman. Despite the
greater particularity implied by the compound with otA,Aa,'saddlei both terms clearly refer to the
same article. According to Maurice's specifications, its contents should include spare bowstrings and it
must be able to hold victuals for at least three to four days, notably emergency rations during combat
operations.s6 This terminology is cognate with Late Latin punga, uniquely attested in a contract from
Ravenna preserved on papyrus and dated 564, in which punga refers to a bag of coins.87 Derivative
terms are found in medieval and modern Greek, and in Romance languages in Romanian and certain Italian dialects (Venetian, Abruzzese, Sardinian).88 Ample correspondents testify indisputably to
a Germanic derivation, including Gothic puggs, Old Norse pungr, Old English pung, OldFrisianpung,
Middle Dutch pong, Middle Low German punge and Old High Cerman pfung. There can be no doubt
that Gothic puggs is the etymon of Latinpunga and Greek nouyylov.8e The priority of the Latin or Greek
form remains uncertairy and interpretation is complicated by the Latin-based argot and Latin-Greek
bilingualism of East Roman armies, but geographical distribution and historical considerations point
to the Balkans as the location of the initial loan and its subsequent transfer to Italy in the course of
sixth-century warfare, a pattern replicated in the documentation of other items of East Roman military
jargon, of both Latin and 'barbariart' derivation.eo It is unclear whether Maurice's (oeAAo)nouyyiov
designates a particular design of saddlebag or merely entails a non-specific nominal borrowing. On the
one hand, the adoption of a Gothic loanword in addition or in preference to the existing and abundant
Latin or Greek vocabulary in this semantic field (e.g. crumena, marsupium, paqoLzr(@oq, saccusloar(x)oq, sacculus, pera/nr1qa, paAavrrcv, yquprEa/yqupala, OuLarcoq, rcr^rqrurcoq) arguably favours specificity
of design or dimension.el On the other hand, Maurice's nouyylov and the punga recorded on the Ravennate papyrus seem sufficiently different in purpose, size and context to allow a generic usage. In either
case, the occurrence of the term inthe Strategiconimplies prior interaction between Roman and Gothic
equestrian culturet which may also account for other items in the inventory of cavalry equipment in
the Strategicon.e2
QLaorciov

Finally, although by no means exclusive to military contexts, QAaorctov occurs in the Strategicon as
the term for a water-bottle or canteen. Maurice mentions this vessel in connection with medical orderlies who attend wounded and unhorsed comrades, but elsewhere he implies that all cavalrymen
86
87
88

Maur. Strat. 1.2.16, 42; 7.A.10.L. 3.


P. ltal.I8.3.3 = CLLAXYll652.3.3 (Ravenna564).
Lexica: TLLX.22640.49-55; Du Cange 1688, 1211; Sophocles1887,913; LBG 1359, s.v. nouyyiov. Studies: Moutsos 1993; Knaw.np.1997.

8e MrnLrscu 1968-9,I 494; 1978, 305; 1978-9,ll53; Kcisr-rR. 1989, 1.33, sst. npugg-s;Kr..l^v,np.2011,269-78.
e0 See discussion by Knannn 2011, 273-5. See further below n. 210.

er
t'

remarks of Kneurn 2011,276-7, who highlights size as a determining factor in the choice of terminology.
E.g.Maur. Strat.1.2.42 specifies that each cavalryman must be equipped with a lasso (A<oq6oorcrcov). The occurrence of this
item alongside a hobble (rdbuAov <Latinpedica) relates to the routine pasturing of army horses (cf. 5.3.4,4.74;7.A13,8.10;
9.3.106-27;11,.2.35), but the use of lassos by Roman cavalry in combat or in pursuit of defeated opponents is also documented; thus at an engagement near Dara in 603 a Roman soldier lassoed and nearly captured Khusro Il (Anon. Guidi "10; Chron.
Seert 2.69). The origin of this practice is uncertain, though Malalas appears to associate it with the Goths: at Chron. 14.23
(Thurn 285.75-8) he reports an action in c.42U2in which a senior Roman officer of Gothic origin 'brandished a lasso in the
Gothic fashion' (ipaora(t rai o6rcragw rcata to lot0Lrcov iOoq) and lassoed (eo6rcrceuoev) his Persian opponent. Cf. also
Olymp. frag. 18 for Goths using lassos in combat. Certainty on this point is elusive, however, insofar as lassos had long
featured in the panoply of other neighbouring peoples with equestrian and pastoral traditions (Yeg. Epit.3.23.3), notably
Huns and Bulgars (e.g. Amm. Marc.31.2.9;Soz. HE 7.26.8;Malalas, Chron.18.21 [366.85-90] > Theoph. Chron.278.10-14), and
in the longer term Parthians (Arr. Parth.frag.20: Suda o 278);Sarmatians (Hdt. 7.85; Paus. 1.21.5; Pomp. MeIa 1.114; Val. Flac.
Arg.6.732-3), Alans (jos. BI 7.7.4), and Ephthalites (ps.-|oshua, Chron. A).
See

65

PHILIPRANCE

if it is not included in the inventory of requisite equipment.e3 Contemporary


evidence suggests that soldiers typically had small wooden canteens.ea The earliest cognate in Greek
is QAaorcrov, recorded by Hesychius in the late fifth century.es Corresponding Late Latinflasco or flasca
is not directly attested before the sixth century.e6 A Germanic etymon is commonly adduced: Gothic
flasco; Old High Germanflaska;OldEnglishflasce. Walde and Hofmann identify a Germanic loan specifically into 'romische Soldatensprache', but the evidence allows for multiple cultural contexts and
possibly multi-stage linguistic influences in different regions from both West and East Germanic.eT
possessed this item, even

CATEGORY

4:

UNCERTAIN GERMANIC DERIVATION

burgus
The term burgus appears to have been applied broadly to any small-scale stronghold, fortle! outpost
or watchtower, both on the limites or along internal road systems prone to banditry, with structural
dimensions and design varying according to period and/or region.es The term is attested epigraphically as ear$ as the 150s, by which date it already enjoyed empire-wide currency.ee The lack of earlier
documentation is probably due to the absence of stone building inscriptions on originally wooden
structures as well as the minimal interest of urban literati. Derivative burgarius, a guard, public servant
or official stationed in aburgus, is documented from 13& and thus pushes back the prehistory of burgus
beyond direct evidence into the first third of the second century, if not earlier.1oo The epigraphic record
of compound quadriburgium is too ambiguous and chronologically uncertain to be of assistance.l0l The
earliest explicit indications of the functions of burgi include monitoring incursions by 'bandits' across
the middle Danube (182-5) and a scouting post at a road;'unction on the desert limes oI Numidia constructed as protection for local travellers (188).102 Thereafter the evidence is intermittent, with increasing

Strat. 2.9.29; dv QAaorioLq; ct.7.A.10.4 for the syncopated form QAaorclv. Inventory of cavalry equipment: Strat. 7.2.lnthis
period, diminutive QAaorciov is otherwise attested only in near-contemporary J. Moschus, Pratum spirituale 163 (PG 87.3
3029D).

Men. Prot. fua1.72.3 (Baldwin) notes that Roman soldiers in the 570s had 'water canteens made of wood' (toiq 0bqo16oq
ayyelote {uA[vo4 oriow); cf. Greg. Mag. Dial. 2.78,'small wooden vessels, which are commonly called. flascones' (lignea oascula, quae ztulgo /lascones oocantur).
95
96

Hesychius, Lex. a7549; ciqupaooaAov'rotriArl { QAaorca.rv. (DAclorcarv 66 io'u ei6og norr;glou (Latte I256)
Cl. flasco in Ennodius, Carm. 2.147.7; Greg. Mag. Dial. 2.18; flasca in Isid. Etym. 20.6.2, who alleges a Greek derivation. See
Sofer 193Q 132-3.
See Du CaNcn 1688,1683; TLL VI.1 876.58-72,

s.oa. flasca, flasco; BnilcH 1913, 6,103; War.nr-Florrr,rer.rx I 513; Mrrinscu 1968-9,
I49O;1978-9,II 50; Casrrrraur 200O 57.
oe Ruccrrno 7895-1987,I 1053-4; Smcr 1897; TLLll2250.77-61; PnrvNrNcr 7940-45 with older bibliography; Baerz 1981; Isaac
7992, 178-86; VIsv 2009. See also McMur.r.nN 7963, 140-51. for the diversity of form, context and terminology of tower-like
structures in Late Antiquity.
AE 1957.279 = 2000.1291 = ILBulg 277 (151-152; Teteven, Moesia inferior); AE 7927.49 = 168 (154-155; Deultum/Krlrdzhali,

Thracia).
The earliest instance of burgarius is CILilI 13795 = ILS 8909 = AE 1895.64 (138; Rakovi(a, Dacia) subsequently CIL lll 13796
= ILS 9180 = AE 1895.65 (140; Rakovitza, Dacia); AE 1977.752 (II-III; Sacidava/Dunireni, Moesia inferior); IGBulg III.2 1690.61
(202; Augusta Traiana/Stara Zagora, Thracia): pougyagic.rv; and possibly 1976.626 (early IV; Tropaeum Traiani/Adamclisi).
See also Cod. Th. 7.1a Q98\; lsid. Etym. 9.4.28.
ClLlll3771b;70691a;10691b (Komdron/Brigetio, Pannonia superior); see also likely or possible reconstructions: AE 1908.53b
(Castra ad Herculem/Pilismar6t, Pannonia superior): q(u)ailrib(urgium); AE "1947.28 (Ulcisia Castra/Szentendre, Pannonia
superior) : (q) uu d (r i)b u(r gium).
AE 1910.145 (with AE 7952 p. 17 n. 15) : RIU 1135 (182-185; Intercisa/Dunaujv6ros, Pannonia inferior): ripam omnem burgis a
solo extructis item praesidi(i)s per loca opportuna ad clandestinos latrunculorum transitus oppositis muniztit (cf. CIL III 3385 : ILS
395 = RIU 1426; ClLl1l10312 = RIU 1131; ILS 8913 = AE 1905.114 - RIU 1136; AE 1998.1057 : RIU Suppl. 159; RIU 1127;1128;
1129; 1130; RIU Suppl. 239). C[ VIII 2495 = AE 7909.73 (188; Calceus Herculis/El Kantara, Numidia): burgum lCommodianumf
slpleculatorium inter duas aias ad salutem commeantium noaa tutelll)a (cf. CIL 11172376 = AE 18955.60; ClLYlIl2494:1L52636).

66

THE PHILOLOGICAL EVIDENCE FOR GERMANI IN THE LATE ROMAN ARMY

frequency in the later 360s and early 370s.103 The potential scale of such structures is indicated by one
report that the entire construction required 48 days.loa The distinctions, if any, between aburgus and
structures variously designated by other terms - turris, castellum, praesidium, centenarium, munimentum,
Qgodqov - remain hazy and it seems safe to assume a degree of synonymy or terminological overlap.10s One can accordingly discern burgi in the non-technical vocabulary of contemporary accounts of
late Roman fortification initiatives along the limes.lo' Typically for a terminus technicus in the military
sphere, burgus does not appear in literary sources until the late fourth/early fifth centuries. The first
instance is in Vegetius, who supplies a definition: castellum paraulum, quemburgum oocant, albeit in the
specific context of a temporary outwork built to assist in the defence of a larger city or fortress.lo7 In
contrast, Orosius seems to envisage primarily a dwelling-place rather than a fortification: crebra per
limitem habitacula constituta burgos tsocant. He also reports a (spurious) etymology of the Burgundians,
conceived as inhabitants of burgi.1o8 Around the same tirne, burgus also first emerges in imperial legislation.l0e The Notitia dignitatum (4001401-c.425) lists the locations of several garrisons in which burgus is
either a component of a specific toponym or a generic term for a fortress.1l0
Late Latin authors exhibit no knowledge of the derivation of burgus or awareness that it was a foreign term - even Orosius' etymologising tale of the ethnogenesis of the Burgundians does not make
burgus a barbarian word or concept. The similarity between burgus and Germanic feminine substantive*burgs (Gothic bailrgs; Old Norse borg;OldHigh Germaru Old Saxort Old English, Old Frisian bzrg)
supplies an attractive etymology. One line of philological scholarship therefore holds that the Romans
formulated burgus from a Germanic etymon at some time before the early second century.11l This view
rests primarily on apparent lexical correspondence and a shared semantic context of fortification. Also
adduced in favour of this model is the allegedly feminine gender of some instances of Latin burgus,
which is explained as a relic of Germanic feminine "burgs. A Germanic derivation is accepted without reservations in some standard etymological lexica.lr2 The thesis is not without difficulties. At a
103

CIL Ylll22629 = ILS 5849 = AE 1889.90 = 1952.15 (184-185; Albulae/Ain Temouchent, Mauretania Caesariensis); C|| Vm
27662(184-792; Bou-Tl6lis, Mauretania Caesariensis); CIL XIII6509 = ILS 2614 (II-III; Schlossau, Germania superior); CIL VI|I
2494=1L52636 (211-217; Calceus Herculis/El Kantara, Numidia); CIL VIII 3 (222-235; Al Jabal al Gharbi, Africa proconsularis); CIL XIII 11976 = AE 2004.1044 (269;Liesenictr, Germania superior); CIL III 3653 = ILS 775 : RIU ZZL = AE 19991264 (327;
Solva/Esztergom, Pannonia superior); A82000.7223 = RIU Suppl. 128 (371; Pone Novata/Visegrdd, Pannonia superior); CIL
III 88 : ILS 773 (371; Umm el-fimal, Arabia); CIL Xm 11538 = ILS 8949 : AE 1993.714: 2002.1051 (371; Mettauertal, Germania
superior); RIU 804 (372;Pone Novata/Visegr6d);
ilI 5670a=lLS774 (367-375; Lauriacum/Ybbs an der Donau, Noricum).

C[

See also the

restoration in CIL III

6159

=7494=ILS770 (c.369; Cius/Girliciu, Moesia inferior). See later: I. Caesarea Maritima

57.4 (c.500): 6 poOgyoq.

r04 CIL III 3653 = ILS 775 = RIIJ 771 : AE 1999.1264

.106

107

(371,; SolvalBsztergom, Pannonia superior): hunc burgum ... a


fundamentis et
construxit ad summam manum operis in diebus XXXXVIII.
See glosses: CGLII426.26: rcuQyoq haec turris,bu<r>gus;I1570.34:burgus turris;lY 27.1:burgos castra (withduplicate instances listed at VI 157). See also the apparent synonymy in two inscriptions from Mauretania Caesariensis commemorating
Commodus' refurbishment of the province's defences c.184-185: CIL VIII 22629 = ILS 5849 : AE 1889.90 = 7gS2.1S (AlbulaeT
Ain Temouchent): burgis noois proaincia munita; CIL VIII 20815 = ILS 396 = AE 1902.220: lgS2.1S (AuzialSour el Ghozlane):
securitati proaincialium suorum consulens turres noaas instituit.
e.g. Valentinian's fortifications along the Rhine in c.369 tn Amm. Marc. 28.2.1. (castra extollens altius et castella turresque adsiduas per habiles locos et opportunos; c|.29.4.1;30.7.6; Diocletian's building programme in Zo s.2.34.1. (noA6or rcai,
eqouqr6rq ra|
nfqyoLs).
Yeg. Epit.4.10.3.

108

Oros.7.32.1J'-12: . . . per castra dispositos lBurgundiones) in magnam coaluisse gentem atque ila etiam nomen ex opere praesumpsisse,
quia crebra per limitem habitacula constituta burgos uulgo uocant .. . . This passage is reproduced almost verbati mby lsid,. Etym.
9.2.99, 4.28; see Sornn 1930 85-6. The same semantic association is found in Eugippius, V. Sea. 4.7: ad secretum habitaculum,
quod burgum lot Burgum?l appellabatur ab accolis. See DrBsrNnnncnn 2OOi., 94-5.
1@ Cod. Th 12.19.2 (400): qui curiae oel collegio
ael burgis ceterisque corporibus ... serztieret; cf. Cod. lust. 1.27.294 (Sg4).
110 Nof' dig. or. 31.63: Burgo Seaeri (Thebaid);
42.28: Burgo Dono, 36: Burgo Nozto (cf. Procop. Aed. 3.6: Bouqlouq v6r1e; both Dacia
ripensis); occ. 33.44: nunc in burgo contra Florentiam, A: ad burgum centenarium (both Valeria). See previously ltin. Anton. \75.4:
Orudisza ad Burgum (Thracia). See below for Sid. Apoll. Carm.22 tit. et passim: Burgus Pontii Leontii (rrrodern Bourg-sur-

Gironde).
111 DuCaNcE"1678,l786;Drc21887,59-6Os.v.borgo;5nncx7897,1066;Ksrdpe1901,347,364-5,386;NonoeN1925,492n.1;Trrrru112

necs1973,24-8. Kr.ucr-Snenoro 1989, 114-15 s.v. Burg offers a convenient summary of the arguments.

Enuour-Mrrt-tz'r 78,'mot 6videmment germanique'.

67

PHILIPRANCE

historical level, the cultural or military dynamic of a loan from Germanic"burgs defies easy explanatiory especially in the context of Roman-Germanic contacts up to the mid second century. The Roman
structures to which Latin burgus was applied, from the 130s at the latest manifest no obvious debt to
contemporary Germanic designs, techniques or capabilities in fortification, certainly none sufficient
to account for the adoption of "burgs in addition or in preference to existing and apparently synonymous Latin terminology (turris, castellum).Indeed, in the semantic sphere of building technology the
traffic of loanwords flowed in the opposite direction.113 Furthermore, Germanic *burgs and its derivatives ordinarily denominate a fortified town or settlemenf of a dimension and character expressed
by terminological equivalence to oppidum, urbs, ciaitas or 7r6LLq, and thence also as a component in
toponyms, rather than a fortlet or outpost on a frontier, though of course linguistic loans can harbour
misconception.lla In any case, the inferred etymological significance of feminine burgus does not survive scrutiny. Apparent usage of the feminine is restricted to a single poem by Sidonius Apollinaris,
writing in the early 460s and thus one of the latest witnesses, compared to the masculine found in all
other instances in which the gender is indicated, including the earliest evidence.lls Sidonius, however,
uniquely applies this term to an entirely different type of edifice, not a frontier outpost, watchtower or
military installatiory but rather a walled country mansiory with magnificent amenities, grounds and
ornamentation, called Burgus Pontii Leontii (modern Bourg-sur-Gironde). In fact, it is clear that Sidonius
conceives Burgus as a proper name and uses the feminine in relation to this grand residence merely by
typological analogy to domus,oillaor urbs, aconvention found elsewhere in his writings (e.9. Carm.18.1;
Aaitacum ... nostram lsc. aillam); ep.8.8.1: tua fsc. oillal ... Taionacus).116
Given these philological and historical considerationg many philologists have long preferred an
alternative derivation from Greek nfq yoq,'tower'. This proposal has much to recommend it: rcriqryog is
semantically closer to burgus, and both are masculine and have a vocalic root, in contrast to feminine
and consonantal*burgs. Lenition of p > b in loans from Greek into Latin, here presumably in a demotic
register, is a phenomenon of which other specimens can be cited (e.g. nugq6q > burrus; nuEog> buxus;
rcaqr;,aooq> carbasus), though all seem to belong to a significantly earlier linguistic horizon. A broad
historico-cultural context for the loan is easily located within the Roman debt to the Greek world in
matters relating to fortifications and poliorcetic technology, as much earlier exemplified by the derivation of Latin turris from tilqrouq, n1qgr.q.117 This thesis does not exclude the possibility of the subsequent
secondary influence of Germanic *burgs on both the meaning and usage of Latin burgus and its derivatives in Romance languages, which by a process of assimilation extended the originally narrow meaning of miqyog > burgus to include also, and later primarily, a settlement.ll8
The evidence does not permit definitive conclusions, but given the existence of an at least equally, if
not more, compelling Greek etymon and the absence of any'Germanic' character, context or rationale
to Latin burgi,we have no option but to treat a Germanic etymology with caution and assign burgus an
uncertain status in this survey.

113 PnNr.rrcx 1940-45,13 n. 2; KuHN 1972, 77-18.


114 PENrsrNcx 1940-45,9-10; NruuaNN 1981.

r15 The feminine occurs solely in Sid. Apoll. Carm.22. prol. 3: burgum tuam ... meam
feci;22.235: burgus placitura petatur. Masculine: CIL Vlll 2495 : AE 1909.73 (188): burgum sfp)eculatorium; CIL III 567Oa = ILS 774 (367-375): hunc burgum; CIL III 88 = ILS
773 (371): fabricatus est burgus; CIL III 3653 : ILS 775 = RIU 771 : AE 1999.1264 (371): hunc burgum; AE 2000.1223 = RIU Suppl.
128 (371): hunc burgum; RIU 804 (372): hunc burgum; Not. dig. 0cc.33.62; ad burgum centenarium.
116 See remarks of PnNNrNcr 7940-45, 10-11. See detailed commentary in DnI.rmv 1993,8-9,60-6L,203, citing other examples.
Delhey (9-12) dates the poem to between46T-7, probably 462-3.
117 The Greek derivation of burgus has a long pedigree stretching back to CasaunoN 1582 Comm. 63, but was first articulated
in detail by Mucr 7897, 113-74 (see also Mucr. 1967, 87-9); similarly TLL 112250.77-19; Bnii*r 1913, 1,5, 17 with 151-3; WerorHorlaaxx 1124, stt. burgus, changed from ed.1 7906,76: endorsing a Germanic etymology; ed.2 1970, 102: offering both derivations. See full argumentation in PruNrcr 7940-45 with other bibliography; endorsed by Bal,lrz 1981,84; Nnunar.rN 1981,
777:18.
118 The latet semantic evolution of bargzs

68

in the Middle Ages is treated in detail in

var.r

Wsnvorr 1965.

THE PHILOLOGICAL EVIDENCE FOR GERMANI IN THE LATE ROMAN ARMY

barditus, barritus

The form, etymology and meaning of this term have been the subject of conflicting opinions since
the late sixteenth century; no claim to a satisfactory resolution of these issues is made here. Ammianus (writing c.390-91) and Vegetius (383-450) report that contemporary Roman infantry favoured a
particular style of battle cry which both authors term barritus.lle According to Ammianuq this thout,
starting off in the very heat of battle as a low murmuring and gradually growing louder, rises in volume, like waves crashing against the rocks'.120 He twice explicitly states that barritus is a foreign word
used by and originating among the barbari: 'the terrifying roaring, which the barbarians call barritus',
and later: from all sides the Romans collectively sounding their war cry, as customary rising from a
low to a louder tone, which in the foreign tongue they name barritus'.121 Lactantius Placidus, probably
writing in the first half of the fifth century, also associates the term barritus with a martial 'howling'
(ululatus) customary among barbari.l22 When Ammianus first introducesbarritus, at least to judge by the
extant portion of his Res gestae, he describes the battle cry employed by the Cornuti and Bracchiati, two
auxilia palatina of Germanic origin. Elsewhere, however, he indicates that this custom was observed
throughout the contemporary Roman army regardless of the ethnicity of the troops, including units
in the eastern empire, which is also the implication of Vegetius.l23Ostensibly, the overall tenor of this
evidence is that barritus was a specific crescendoing war cry, originating in a Germanic military practice, which had entered the Roman army through its prevalence among auxilia palatina, elite infantry
units typically recruited, at least originally, from diverse Germanic population-groups living east of
the Rhine from the late third to later fourth centuries.l2a
While there is little reason to doubt the Germanic origin of the practice, the etymological derivation
and the correct form and orthography of this term have long been disputed. The controversy hinges on
the relationship between the barritus documented in late fourth- and fifth-century texts and a practice
reported by Tacitus three centuries earlier. Inhis Germania, complled c.98, Tacitus writes:
'They have also these songs, by the recital of which - "barditt.ts" they call it (carmina, quorum relatu,
quem barditum oocant), they enflame their spirits and from the singing itself (jpso cantu) they augur the
outcome of the impending battle. For they inspire or indeed feel dread in proportion to the resounding
of their line, and it seems not so much an articulate sound as a concerted expression of valour. They
aspire particularly to a harshness of tone and an irregular roar, putting their shields to their mouths,
so that by reverberation it may swell to a fuller and deeper sound,.12s
Accordingly, the Germani chant'songs' (carmina) before battle, which are reportedly amplified and
reverberate against the backs of their shieldE with the aim of both affirming their collective morale
and intimidating the enemy, and also with an alleged augural purpose. Tacitus states that this particular manner of performance or recital (relatus) is called barditus, which thus denominates the mode
rle Ammianus

76.72.43;27.13.15;26.7.77;31.7.11

withDrloNca1972,255;ozvBosvretal.lggl,2Tl.ThecompletionofAmmianus'

is securely dated to c.390-1, see Mer.rHaws 2007,12-27. Vegetius: Epit.3.lB.9.


120 Amm. Marc. 16.12.43: qui clamor ipso
ferztore certaminum a tenui susurro exoriens paulatimque adulescens ritu extollitur fluctuum
cautibus illisorum. Cf. similarly 31".7.11: a minore solita ad maiorem protolli. Cf. Yeg. Epit. 3.18.9: Clamor autem quem barritum
aocant, whose injunction against using battle cries until immediately prior to engaging with enemy is consistent with Ammianus' ipso feraore cerlaminum.
121 Amm. Marc.26.7.17: terrifico
fremitu, quem barbari dicunt barritum; 37.7."11 Et Romani quidem ooce undique Martia concinentes, a
minore solita ad maiorem protolli, quam lsc. ztox Martial gentilitate appetlant barritum. See TLL V17879.46-5g for the meaning of
gentilitate as 'sermone barbaro'.
r22 Lactant. PI ac. ln Stat. Theb. 4.394: ululatum: quod
Amtzones cum fnitimis bellare dicuntur, quorum ululatus barritus, id est barbarus
ritus. Cl. ibid.5.729.
Res gestae

123 CornutiandBracchiati:Amm.Marc.76.12.43;generally2-1..73.75;26.7.12;37.7.11;cf.yeg.Epit.3.7g.9.
124 Ar.r6rot 7959, 174-5, followed
by Mrrr.rrn 1996, 10L n. 2, over-interprets Amm. Marc. 76.12.43 in deducing that the

Cornuti

specifically were responsible for introducing this practice into the Roman army. On auxilia palatina see most recently Snrrosr 2004b with older bibliography; CoLoMso 2008, esp.136-9.

12s Tac.Germ.3.1-2:Suntillishaecquoquecarminaquorumrelatu,quembarditumaocant,accenduntanimosfuturaequepugnaefortunam
ipso cantu augurantur; terrent enim trepidantzte, prout sonuit acies, nec tam oocis ille quam airtutis concentus rsideniur. Adfectatur
praecipue asperitas soni et fractum mutmur, obiectis ad os scutis, quo plenior et graoior aox repercussu intumescat.

69

PHILIP RANCE

of chanting or intonation, not specific songs or war cries. Customary'singing', broadly construed, is
sporadically reported in other accounts of Germanic warriors.l26 This passage has generated a large
bibliography. The interpretative difficulties are compounded by the variant reading baritum rather
than barditum in inferior manuscripts, which has tempted some editors and commentators to harmonise Tacitus' text with barritus transmitted in late Roman sources, although this recourse does not in
itself resolve the wider etymological and historical difficulties.l2T Furthermorg it is not entirely clear
whether Tacitus' wording quem ... aocant is intended to introduce a Germanic or a Latin term. Beyond
the issue of oariae lectiones, older philological scholarship concerned itself primarily with attempts to
determine the precise nature and significance of the martial practice Tacitus describes, and in particular whether he refers to war songs or a battle cry. More recent studies have demonstrated the
Greco-Roman acoustic and psychological conceptions under$ing Tacitus' accounf to a point that potentially casts doubt at least on his interpretation of this custom, if not its authenticity.l2s If barditus is
what Tacitus wrote, its etymology continues to defy corrrpelling explanation. When not demonstrably
false, none of the several hypotheses is wholly convincing, but certain propositions have periodically
found favour. A posited connection with Old Norse bardi,'shield', envisages a 'shield song', although
this term is a hapax attested in a late gloss. A proposed link with Latinbardus, 'bard, singer', a Celtic
loanword, would in turn raise questions concerning Tacitus' sources of information and the possibility of Germanic-Celtic cultural interaction. Alternatively, proponents of Tacitean bar(r)itus, principally
Muctu have adduced Swiss dialect barren,'to cavort raucouslyl Gebarr,'uproar, din. Nothing new can
be added to this controversy here.12e
The late Roman evidence is further complicated by the prior existence of a Latin noun barritus (<
barrio,-ire, 'to roar, to bellow'), meaning a'bellowing' or'roaring', originally associated with the 'trumpeting' of an elephant and derived from Latin barrus, a Sanskrit loanword for this anima1.130 Indeed,
Vegetius elsewhere inhis Epitoma uses barritus in this alternative sense.131 How can barritus be both a
Germanic loanword and an organic development within Latin? Some scholars have doubted or disallowed a connection between the historically distant reports of Tacitus and Ammianus, emphasising
the dissimilarity of the practices described - a style of pre-battle augural chanting in contrast to a specific war cry during combat - and the different terms used - barditus versus barritus.132 Nevertheless,
Ammianus indicates thaf in his opinion, the war cry he termsbarritus is'barbarian', both as a practice
and, explicitly, in name. Even accepting that Ammianus' Latinity elsewhere exhibits self-conscious
Tacitean echoes, his usage of barritus for a contemporary war cry is confirmed by Vegetius and Lactantius Placidus and cannot be discounted as mere rhetorical erudition.133 Another strand of scholarship
has therefore found the commonalities of the first- and fourth-century evidence too great to be plau-

"'

E.g. Tac. Hist. 2.22 (cantu truci); 4.18.3 (oirorum cantu, feminarum ululatu sonuit acies); Ann. 4.47.3 (cantuum ... tumultu); Plut.
Mar. 19.-4; Lib. Or. 18.51 (natav[(ovtcq).
127 For aariaelectiones seeRosrr.rsoN 1935,123,276app. crit. SeeTLL111750.82-4:barditus.
1T For a summary of older scholarship see Bnucxnnn 1907, 65-8. See the critical review of subsequent studies in LuNo 1991,
2028-31. Commentaries: Muctt 7967, 77-80; LuNo (1988) 72-3, 118-19; Prnr. 1990, 138-9; Rrves 1999, 123-4. An alternative hypothesis devised by Lumo 7985, 263-70 champions the variant reading baritum, which Lund interprets as an otherwise
unattested Grecism (< paqUtov(ov)) and thus purely descriptive of a musical mode; the argument is ingenious but, in my

view, unpersuasive.
12e Concerning the etymology of barditus, the main lines of argument and rebuttal had been set out by the beginning of the
twentienth century. See the review by Bnucxrvrn 1907,65-8; supplemented by Mucu 1967,77-80. For other opinions see also
the selective resum6 by Bncx1976, although he does not always correctly assign hypotheses to their originators.

rs
'"'

The earliest instance occurs in the second century: Apul. Flor. 77; seeTLLIl1756.75-1757.9:barritus;1757.34-48:barrus.
V"g. Epit.3.24.5: Elephanti in proelis magnitudine corporum, barritus horrore, formae ipsius nooitate homines equosque conturbant,
'Elephants in battle throw men and horses into panic on account of the size of their bodies, the horror of their trumpeting
and the unfamiliarity of their very form'. Here barritus clearly has a different meaning to Epit. 3.18.9, where it is a contemporary term for a specific battle cry.
132 See recently e.g. Nrcasrn 1998, 108-10, who accepts that the late Roman barritus is a Germanic custom, and
possibly even a
Germanic word, but is sceptical of a connection to Tacitus'earlier report.
133 Ammianus' imitation of or allusions to Tacitus: Sannau 7978, esp.15-19, 101-10; Krrrv 2008, 76:24,175-8,188-90,200-201",21420.

70

THE PHILOLOGICAL EVIDENCE FOR GERMANI IN THE LATE ROMAN ARMY

sibly explained by coincidence alone - Germanic warriors arrayed on the battlefield; martial clamour,
whatever its precise occasiory form or objective; and the close lexical resemblance of barditus-barritus.
Accordingly, they interpret the testimony of Tacitus and Ammianus as evidence of long-term continuity in Germanic military culture. While aspects of this thesis remain uncertain, the most plausible
explanation is the deformation of a Germanic loanbarditus either by assimilation to Latinbarritus or,
more plausibly, owing to their popular and perhaps humorous association within the argot of Roman
soldiers, who identified a Germanic term for a war cry with their own word for bellicose'trumpeting'.
Late Roman authors were aware of the non-Roman origin and designation of this war cry, but apparently ignorant of the etymological process.l3a If this analysis is correct, the original geographical and
cultural milieu for this (albeit modified) terminological loan for an imported Germanic practice would
seem to be the armed forces of the western empire in the late third and/or early fourth centuries.
*caulis

This lexical puzzle concerns an apparently corrupted passage of Vegetius' Epitoma. The text refers to
the oinea, a kind of mantlet or moveable shed constructed from light timber and wicker and used to
shelter sappers and assault-troops during siege operations: Vineas dixerunt aeteres qufrs nunc militqri
barbaricoque usu cqucias aocant,'The ancients called "yirtes" what are now termed cauciae in military
and barbarian parlance'. The manuscripts offer variant readings: caucias b, cautibus e, cautiusT; caucius
Z (sigla ed. Reeve).135 The three modern critical editions do not offer a decisive solution. Unconvinced
by the conflicting testimony of the manuscript prototypes, Lang (1885) substituted causias, an editorial
conjecture which can be traced back at least as far as the early sixteenth century. This seeks to identify
Vegetius' term as causia (< rcavola), a type of Macedonian broad-brimmed sunhat, otherwise attested
in Latin infrequently between the ear$ second century BC and first century AD. This interpretation
has the attraction of offering a known Latin word, and one which supplies a plausible terminological
analogy for a protective covering, but causia lacks manuscript authority and it might be reasonably
doubted whether this recherch6 Grecism would have found its way into or could be described as
militari barbaricoque usu.B6 }nnerfors (1995) likewise favours causias.l37 Reeve (2004) prefers caucias on
codicological grounds, even if the meaning and etymology of the consequent hapax thus remain unresolved.138

Vegetius implies that this term is current in both the contemporary slang of Roman soldiers and
among unspecified barbarians; he does not explicitly state that the word is of non-Roman origin. Kempf
tentatively proposed that the received text might conceal a word with a Germanic or possibly Celtic
root. Of the variant readings listed in Lang's apparatus criticus (1885), Kempf selected cautias (P). He
drew attention to potential correspondents in Germanic languages: Germanic kot, kote, which Kempf
defined by Latin tasa'; Old English cyte; Old Norse cytia (both of which he derived from a common

root*cutia), equivalent to Latin tasella'.13e Subsequenf and better-informed, etymological scholarship


clarifies these cognate forms in North and West Germanic languages, some of which provide broad
conceptual parallels to a oinea: Old Norse kot, kytia,'small cottage'; Old English: cot, cote, cyte, 'cottage,
house, chamber, den'; Norwegiankot4ta, 'wooden hut made from branches'; Low German (borrowed
into Modern High German) kot, kote,'shed, stable, hut, stalf cottage'; Middle Dutch cot, cote, 'cave,lair,
134 SeeInu1897;Knv.w1901.,350,372,386-7;Bnucxurn1907,esp.66-8;Gnossr7920,250;HorruaNN"1969-70,1135-7;KuttN1972,
43-4; Bncx 1976; SrnrNcnn 2002; Sprloet 2004a, 1-10-13. Mucn 7967, 77-8Q although preferring to rcad bar(r)itus in Tac. Cerm.
3.1, also

identifies historico-cultural continuity with the late Roman evidence.

V"g. Epit.4.75.7. The evidence lor the ainea is assembled by LENorr 1983,1.39-4L.
"t
136 Larvc 1885, 137-8 app. crit. See TLL lll 659.37-47, sn. causea (-la), which cites Veg. Epit.4.15.L with variant
does not acknowledge the occurrence of

readings. Lang

this emendation in some previous editions, of which the earliest appears to be C.

Wechel (Paris 1532) 138.

13' Orgurrpons 1995,214.292.


r38 Rnrvn 2004,733 app. crit., with remarks on codices at xxviii-xxix.
r3e Ksr4pp 190L,348,365 with n. 3.

77

PHILIP RANCE

stable, den'. There is no attested correspondent in East Germanic.lao Philologically, however attractive,

this proposal cannot proceed beyond conjecture, especially given the uncertainty of the Latin text. In
particular, Reeve has since demonstrated that the reading cautias uniquely found in codex P is itself
later conjecture and thus not an authentic witness to Vegetius' text (except insofar as it might be a lucky
guess).l4r Nevertheless, the component caut- in variant readings in e and T continues to provide hope
for any who wish to pursue Kempf's hypothesis.la2
In Roman sources of the Principate and Late Empire, Germanic peoples, especially in western Europe, notoriously lack capabilities in siegecraft and particularly poliorcetic engines.la3 It is possible,
however, that Roman writers overstated or generalised this deficiency; certainly exceptional episodes,
notably the Gothic sieges of cites in the Balkans in the 250s and 260s, suggest that technological knowledge varied between different Germanic peoples and/or periods.raa The siege machinery reportedly used by Germanic peoples includes mobile mantlets, typically covered with flame-repellent oxhides.las Although the evidence is slim, these were hardly sophisticated structures; indeed their very
simplicity arguably enhances the likelihood that such a device was used by or particularly associated
with 'barbarians'. More generally, historical parallels may be drawn with other instances in which
technologically advanced late Roman and Byzantine armies adopted similarly simple contraptions,
sometimes together with their foreign terminology, from peoples traditionally credited with inferior
or no expertise in siegecraft. Two examples suffice. First, during the Roman siege of Petra in Lazica in
552, where the gradient prevented the use of wheeled battering rams, the Romans were reportedly at a
loss how to proceed until some allied Sabir Huns taught them how to improvise skin-covered wicker
shedg designed to support a battering ram but light enough to be carried on the shoulders of the men
within.la6 Second, at a later date, tenth- and eleventh-century Byzantine authors employ the term Aaioa
(or A6oa) to denote a light portable hut-like shelter with a sharply peaked roof, which could quickly
be constructed in situ from vine stalks or woven branches as protection for sappers, and thus a structure corresponding exactly to Roman rineae. The word Ldtoahas been identified as a loan from Slavic
languages, in which the core meaning is 'fencing', specifically of wickerwork.laT Correspondingly, if
East Roman lByzantine military engineers could take lessons from Sabirs and Slavs on the construction of such mantlets, it is not beyond imagining that Vegetius refers to a loanwordbarbarico usu for a
borrowed 'barbaiar{ design for a oinea, but whether or not the etymology of this garbled reading is
Germanic currently defies demonstration.
hornatores

Unique testimony to hornatores is found in a gloss in the so-called Glossae affatim, of which the oldest
manuscripts date to the late eighth/early ninth century: Liticin<e>s: hornatores cornices aut cornicines.If
authentic, this word apparently designates a type of trumpeter or bugler.la8 Kempf conjectured a hybrid formation within Roman sermo castrensrs, whereby Germanic Horn penetrated Latin Q "hornum)
lao PoxorNY
141

7959, 393-8, 586-7.

Both Lang and Onnerfors admit readings from codex P (codex Leiden Periz. F 17) into their apparatus criticus. From Lang
the variant cauflas also found its way into TLL III 659.38-9. See general remarks on P in RrrvE 2004, xxviii-xxb<, lvi.
142 Since Kempf's analysis, the question has attracted little interest. Hsnaeus 1902,264 acknowledges the superior authority
of cautibus over cautias, but his remark'was auf cautiaos fiihren wiirde' is of no great assistance. Moscr Sassr 1983, 125-6
prelerc causias, with the meaning of a sunhat. P6nrz Cesrno 2005,82-3 appears to accept a Germanic etymology, but without
argumentation.
143 See convenient summaries in TsoursoN 1965, 84-6; Error.r 1996,82-6.
144 See the siege machinery described in Dexippus, FGIH 100 frag. 27, 29; Eusebius, FGrH 101 fuag. 2; Zos. 7.43.1.
'* E.B. Dexippug FGrH 100 frag. 27.4: Gothic siege of Philippopolis in 250. See also probably Euseb. FGrH 101 ftag. 2.4: an
unnamed city in Macedonia. Cf. Priscus fuag.6.2 (Blockley) for similar protective screens used by Attila's army at the siege
of Naissus in 442.
ra6 Procop. BG 4.1L.27-34. Shortly afterwards, another group of Sabirs constructed the same devices for the Persians: 8.14.4-5.
147 Evidence and discussion: McGrnn 7997, 135-8; Surr-rvau 2000,175-7.
r48 CGL lV 534.37, with codices of the Glossae
ffitim identifred at lY 471..

72

THE PHILOLOGICAL EVIDENCE FOR GERMANI IN THE LATE ROMAN ARMY

and received the termination -ator, by analogy to aeneator and bucinator.rae Subsequent studies have
found this etymology overly bold. Even before Kempf's interpretation, Goetz had suggested that the
reading hornatores was in fact a corruptionof aen(e)atores.Thishas subsequently become the prevailing

view, supported by similar corruption traceable in corresponding glosses.lso In these circumstances,


nothing further can be said with confidence. Historical sources are of no assistance. With some imagination one might wish to discern a 'Germanic'-style horn in Vegetius' explanation of the various
musical instruments (cornu, tubn, bucina) traditionally used by the Roman army, where he identifies a
cornu as the horn of an aurochs, a beast long associated with the Germani.lsl This particular passage
is beset with interpretative difficulties, however, and, more generally, the design and construction of
these instruments, and even the specificity of their terminology, are less certain than some modern
literature allows.ls2
roOA\oq, roOASov
The word to0Aboq, later to0A6ov, is a Late Latin terminus technicus for the baggage-train of the army in
the broadest sense. Not directly attested in Latin sourcet it is documented in Greek transliteration in
connection with the East Roman army from the sixth century and became the standard terminology
in Byzantine military literature. The earliest instance is possibly in Urbicius' Tacticon, written in the
reign of Anastasius (491-518), although it is uncertain whether this reference was an original feature
or later interpolation.ls3 Otherwise the term first occurs in Maurice's Strategicon, by which date it had
become a standard component of contemporary military jargon. In his introductory list of terminological definitions, Maurice explains: 'the'touldos' is the baggage of the soldiers, that is servants and pack
animals and the other beastsi though elsewhere it is clear that this term essentially refers to a train
of ox-drawn wagons.lsa The earliest occurrence in a non-technical work is the Historiae of Theophylact
Simocatta, writing c.628-30, who indicates that this word belongs to the'native tongue' of the Romanq
a circumlocutionary phrase he employs whenever resorting to a Latinism stylistically unacceptable in

lae See Knurr 1901, 352,378,387. See also


NrrosnNraNN 1905, 19.
r50 See the tentative emendation offered
by G. Goeb in CGL VI.1 (1899) 650 s.v. Liticen,'hornatores (aeneatores?)'. See thus Hrnarus 1902, 273;TLLY[.3 2972.3-7; Moscr Sassr 1983, 135; P6nnz Casrro 2005, 87 n. 25. The process of corruption aeneatores >
hornatores may be traced in parallel textual corruption in other glosses: CGL II 12 Aen(e)aior;V 185.32: Cornicenes: orniatores
qui de cornu tubas habent, with the interpretation of Lrxosay 1926-37, I 148 Ns 2252:'orniatores (aenea-)'.
'u' V"8. Epit.3.5.6: cornu quod ex uris agrestibus, dlgento nexum, temperato arte spiritu quem canentis flatus emittet auditur, ,the cornu
is that from the wild aurochs, bound in silver, which when modulated by skilful breathing emits a resounding blast,. On
the aurochs (urus) and its horns: Caesar, BG 6.28; Pliny, HN 8.15.15 (38). Maurice, Strat. 12.8.16.4 also refers to i towhorn
(taugda) used by Roman infantry to issue signals. The attempt of Marranas 2001 ,96 to identily taugia here as an ox-hide
drum does not convince.
1s2 The literary, epigraphic
and artistic evidence for musical instruments used by the Roman army is ambiguous or contradictory in all periods. See Spsrosl, 1975; Banror.r 1987; Mruccr 1982; Zrctrowsxt 1999; 2002.
153 Urbicius, Tacticon 11. The
reference occurs as a gloss in the chapter-heading to $11: neqi trle ,rriv oreuoe6gov noqreiaq, toO
vuv Aeyoprivou toilA6ou, ov bei rcLveio0at rcaT a ndvte olrlprara, 'Concerning the march of the baggage-train, which is now
called the touldoslon, which should be moved in accordance with one of five arrangements'. fhe chipter- and sub-headings
that unevenly subdivide the received text of the Tacticon enhance its utility as a 'dictionary of the infantry phalanx, and
may be part of Urbicius' original scheme, but there are reasons to suspect their authenticity, especially given evidence
that a'new edition' of Urbicius' text was produced during the reign of Leo VI (886-911). Such ireadings are not a feature of
Urbicius' model, Arrian's Ars tactica, or any other surviving specimen of this originally late Helleniitic sub-genre. Nor is
Urbicius' Epitedeuma, to which the Tacticonwasformerly conjoined, subdivided in this way, which allows for tlie possibility
that the headings were inserted into the Tacticon after these two texts were separated by (probably) ninth-/tenih-century
Byzantine editors. Futthermore, some of the chapter- and sub-headings have an interrogative or catechistic format, otherwise unknown in classical or late antique military writing, but which finds parallels in ihe Mlddle Byzantine genre, notably Leo's Problemata, written in the 890s. See further ReNcr forthcoming.
15a Drrrqrs 1987,552, index
graec. s.v. to0A6oq. The formal definition is given at Strat.7.3.39-40: Kai- to0Aboq iotw rl arroorceul
td;v otqratrr^rtri-rv, ,routiotr.v, naibeq tr rcai 0no(uyLa rai ra Aotna (G:a.

73

PHILIPRANCE

origin of this term is obscure. Du Cange posited Vulgar Latin"tuldum,but refrains


from further etymologising.ls6 Collinet proposed a Germanic derivation on the basis of Old High Gertman Tuld,'Jahrmarkt', an annual fair (> modern Bayerisch and Austrian dialect Dult; Swiss German
Tult; cf. Gothic dulps,'festival'), though the etymological and semantic links are far from compelling.lsT
Dain rejected Collinet's hypothesis in favour of a Late Latin derivation from posited *tultum (or "tultus
or *tuldus), formulated from tuli, the perfect of fero, apparently with the sense of 'the thing that is carried away', the'load' or'burden'. This proposal is in turn morphologically problematic.ls8 Finally, and
entirely overlooked by subsequent scholarship, Kahane and Kahane posited a Gothic etyrrron *tuld,
modified through Balkan Latin. They contend that this term originally referred to the coarse protective matting which Maurice specifies should cover the rear of each wagory but later came metonymically to denominate the wagon-train itself. In this regard they allege a correspondent in Ibero-Romance
nautical jargontoldo,'awrtirtg| as well as other parallels in Romance languages associated with the concept of sheltering or screening. Since Maurice here, as always elsewhere, terms this matting rctAlrcov (<
Latin cilicium) and it is hardly a crucial or defining characteristic of the baggage-trairu this convoluted
line of reasoning is unpersuasive.lse All that can be said with confidence is that a Germanic derivation
remains unproven and, on current evidence, not especially likely.
Kunstprosa.lss The

caput porci, caput porcinum

Finally, although not strictly a loanword, or even an exact translation or calque, I include in this category of uncertain or doubtful linguistic borrowings the Latin phrase caput porcilporcinum, 'pig's headi
a slang expression which Roman soldiers applied to a specific tactical formation more correctly termed
a cuneus,literally'wedge'. Some scholars have wished to discern in this apparently descriptive label a
Germanic antecedent, as evidenced much later in Old Norse as sts{nfylking, 'swine-array', apparently a
similarly'wedge-like' battle-line.
It is first necessary to clarify the usage of cuneus, a designation employed since the Republic to describe deployments of both infantry and cavalry, both Roman and non-Roman, with a meaning and
nuance that varies according to author, period, context and/or genre. Modern historians have sought
to distinguish different technical generic and metaphorical usages; space permits only a brief delineation here.160 I have excluded any treatment ol cuneus and its cognates used in relation to cavalry units,
deployment and tacticg as having no relevance to the present discussion.l6l It is now generally acknowledged that in historical writing cuneus often signifies little more than a non-linear combat formation, perhaps varying in detailed configuration but typically a deep and dense infantry deploymenf
and in some instances the term conveys an even broader conception of a generic 'unit' or 'troop'.162 In
relation to the infantry of non-Roman peoples, Roman authors describe Germanic armies drawn up
in cuneos or cuneatim, which may have no more specific or technical meaning than'massesi'bands' or
'groupings' determined by kinship or tribal filiation. Nevertheless, in combat both military and societal factors would have required the more experienced and better-equipped warriors - nobles with
their personal retinues or warbands - to be placed at the forefront of a combat formation, and these
warriorg relatively few in number, provided a tactical spearhead for the majority of the host levied
1ss Theoph. Sim. Hlsf. 2.4.7:

rqv te lleqorrrlv arroorceurlv .. ., {v oivqOeq ?c.-rpaiotg t11 dnrlrogiqp Qrovr,l toOAbov arrorcaAeiv,
'the Persian baggage, which Romans are accustomed to call touldon in their native tongue'. For the 'ancestral' or 'native'
tongue cf . also 2.15.9 ; 6.7.9, 9.14; 7.2.6, w ith Bl^t ow tN 197 7; Whitby 1982.
156 Du CaNcr 1678,Y111205.
157 Cor-r-rNrr 1930. See Kr.uce-Ssneor,o 1989,159 sst. Dult.
1s8 Darlv 1950; followed by MrHi.rscu 1968-9,1495. See also Srnrurn 1988,17,'Lat. tultum, toltum (tollo)'.
15e KanaNr-Kauar*n 1961,473-5. The passage in question is Maur. 9trat.72.8.78.6-9 (cf. rctAinov at10.3.9;12.8.6.6,27.74)
160 Delnntcx 1920-28,1133-6,43-4; Guuoer 7939,754-65;Ldrrarrrnnr 194Q 15-31; Nrcasln 1998,"110-12; Sw:iNNe 2004,274--17,368-70
(to be read with caution); WHrrlrn 2004, 327-2, 342-50; JaNNrano 2004b.
161 See RaNce 2004a, 120-25 with bibliography, to be supplemented with Lswtr 2003; BnrNNaN 2007,214-"17.
162 See Gunml 7939, esp.158-9; LaMMsnr 194Q esp. 30-1; Wunerrn 2004,342; RaNce 2004a, 123-4; jervNrano 2004b,7004-72.

74

THE PHILOLOGICAL EVIDENCE FOR GERMANI IN THE LATE ROMAN ARMY

for the occasion from outside this'professional' group.163 With regard to Roman armies, the picture
is more complicated. In some instances, cuneus simply indicates dense or columnar deployments that
were in no sense 'wedge-shaped'.t64 By extension, cuneus could also mean a generic 'troop' or 'band'
of any type number or formation.l6s For Roman tactical writers and military antiquarians, however,
cuneus could denote a specific close-order infantry deployment.l66 This more restricted technical usage
of cuneus may also be discerned in a relatively small number of historical accounts.167 In general terms,
this cuneus was dense and in some way narrow-fronted with the objective of breaking the enemy line,
but the frustratingly elusive nature of the evidence has allowed widely differing interpretations of its
origin, structure and operation.168
In the context of a specific Roman infantry formation termed a cuneus, both Vegetius (writing 383450) and Ammianus (published c.390-91) report that this deployment was also called a'pig's head' in
the argot of contemporary Roman soldiers. Vegetius provides a definition:
'A "wedge" (cuneus) is the name for a mass of infantry, attached to the line, narrower at the front,
thereafter broader, which advances and breaks through the opposing ranks, because missiles are discharged by a larger number of men into one position. The soldiers call this form a "pig's head" (caput
porci)'.16e

The precise configuration of troops described by Vegetius remains disputed, but it is generally
recognised that in this section of his Epitoma he drew on much older source materiaf probably originating in Cato the Elder's lost De re militari, which in turn synthesised Greek tactical theory for Roman
readers.170 Vegetius' supplementary remark on contemporary soldiers' slang is corroborated by Ammianus, who reports an encounter between the army of Constantius II and the Limigantes, a subject
people of the Sarmatians, on the middle Danube in 359. Although the historian's florid rhetorical language conveys, at best, only a general sense of the course of events, he includes the detail that when the
enemy launched a massed assault on the emperor/ the Roman army, 'ending in a narrow front, which
arrangement the soldiers' simple parlance terms 'h pig's head" (caput porcinum), scattered them with
a fierce charge'.171 There can be no doubt that caput porci(num) is an authentic specimen of late fourthcentury sermo castrensls. This popular designation presumably reflected a formal resemblance between
the infantry deployment and the snout of a pig or boar. Neither author indicates or implies that this
terminology had a non-Roman origin.

163 The classic statement is Tac. Germ. 6: acies per cuneos


componitur; cf. Ann.2.45; Hist. 4.16,5.16; also Hisf. 4.20.3 for a cuneus of
Batavians long in Roman service. Amm. Marc.16.72.20twice reports Alamanni in the 350s deployed lz cuneos... stetere cuneafl. See, however, Amm. Marc .77.2.7: Francorum oalidissimos cuneos in sexcentis oelitibus, apparently generic 'bands' of around
600 lightly-armed Frankish raiders. Agathias' description (Hrs t.2.7.8-9) ol the giant Frankish-Alamannic 'wedge' (dprpoAov)
at Casilinum in 554 is problematig see below nn.173-4. For differing assessments of Germanic cunei see Delbriick 7920-28,
ll33-5,43-4; Gundel 1937,11-18; Guuorr 7939,155-8,160-64;Ldv,v.nnr 1940, esp. 28-1; Mucu1967,150-1; Gornswonruv 1996,
49-51; Wurrlen 2004,1346; Rarvco 2004b,290-95; faNxrano 2004b 7002-4,7007-11.

8.8.Amm.Marc.77.12.9:iunctisdensiuscuneis;3L.9.3:congregatosincuneos,inbothcaseapparentlyreferringtogenericcloseorder deployment or compact columns. For other instances and discussion: jaNNra.no 2004b,7003-6.
'uu E.8. Amm. Marc. 16.11.15; 25.6.7; 37.76.4; Claud. Epithal. Pallad. et CeL 86-7; Jer. Vulg., ludices 9.37; 1. 1am.13.17. Jerome renders
as cuneus theHebrew word aalt, meaning a 'band' or 'comp any'; cf. the Septuagint's similarly neutral agpj in both passages.
166 Cf.Cato,Deremil.frag.ll,;Gell.NA10.9.1;Veg.Epit.7.26.6;3.17.4.8,18.6,79.4-7;possiblyMauu..Strat.l2.A.7.24taAcy6peva

'*

triv ne(6v rouvia.

167 The relevant passages are assembled and discussed


by JANNTARD 2004b,101,2-27.
168 See Drr.nniicx 1920-28,1133-6, 43-4; Lanrrarnr 1940, 15:27;
Wurrr.nn 2004, 321-2,342-50;lrNNrARD 2004b, esp.1021-30.
'u' V"g. Epit.3.79.6: Cuneus dicitur multitudo peditum quae iuncta cum acie primo angustior deinde latior procedit et adaersariorum
ordines rumpit, quia a pluribus in unum locum tela mittunter; quam rem milites nominant caput porcinum.
lro It has long been accepted that material on tactical formations in Epitoma 3.77 and 19 derives indirectly from Cato's De re
militari, written in the first half of the second century BC, with Frontinus' lost work of military theory, written c.8 4-96, as a
likely vehicle of transmission: see Sar.ropn 1929; ScHrNx 1930,54-55,60-61; NEuuaNo 1965, 1009-1018; Mrr.Nrn 1996, xvii-xxiii,
100 n. 3, 101 n. 1; Raucn 2004a, 98-9. For Cato's De re militari: Nan 1927; for its place in the Latin tradition see esp. Wunnr.nn

2004,337-40.
17r Amm. Marc. 17.13.9: desinente in angustum
fronte, quem habitum caput porci simplicitas militais appellat, impetu disiecit ardenti.

75

PHILIPRANCE

The only further information of potential relevance is found in Agathias' Historiae, written two centuries later in the 570s, which provides an extended account of the battle of Casilinum in 554. Agathias
reports that the Frankish-Alemannic army deployed in a single giant hollow'wedge':
'The shape of their battle line was like a wedge (EprpoAov), for it resembled the letter delta, and the
front part, inasmuch as it terminated in a narrow poin! was covered over and compact by having been
fenced about on all sides with shield+ so that you might even say that they had given to their array the
appearance of a pig's head (ouoq rceQaLrlt)'.172

The significance of this passage to the present study is by no means clear. Unlike Vegetius and
Ammianus, Agathias does not claim to employ contemporary military terminology and ostensibly his
long-winded allusion to a pig's head appears to be no more than an impressionistic simile, although
the coincidence of a porcine allusion is striking. Interpretation is further complicated by the propensity of Agathiag peculiar among late classicizing historians, to lace his narrative with concepts and
terminology drawn from Greek tactical manuals, allowing for the possibility that his highly literal and
somewhat geometric description of the Frankish-Alemannic wedge has been coloured by his familiarity with this genre.173 Furthermore, scholarship has cast doubt on the reliability or historicity of other
aspects of Agathias"battle-piece'.174

Against this background, Miillenhoff first drew attention to the occurrence of the compound noun
sztinfylking or 'swine-array, swine-deployment' in Old Notse.17s The apparent terminological correspondence prompted Kempf to offer the tentative speculation that both this 'wedge-shaped' infantry
deployment and the Late Latin expression caput porci(num)had a Germanic origin.176 Subsequent studies of the Old Norse evidence have established that soinfylking and the cognate verb soinfylkja,'to deploy in the swine-array',refer to a deep infantry deployment which was fronted by a projecting'snout'

(Old Norse rani) of well-shielded 6lite warriors. It is disputed whether the verbal expression fylkja
hamalt is synonymous with soinfylkja.It is important to clarify that both soinfylking and sainfylkja occur/
rather infrequently, in various genres of medieval Scandinavian literature - saga, semi-historical narrative, romance, speculum regis - in no instance predating the late twelfth century.177 The few studies
of late Roman and/or early medieval warfare to discuss this question have expressed varying degrees
of acceptance or denial concerning the plausibility of a linguistic and/or cultural connection between
soinfylking and caput porci(num) and its potential implications, although for the most part without citing the Scandinavian evidence or engaging with its linguistic and textual difficulties.lTs The problems
of interpretation have been exacerbated by the fact that, outside the philological expertise required to
read Old Norse texts, the only source at the disposal of non-specialists has long been the Gesta Danorum
of Saxo Grammaticus, a classicizing and semi-mythical history of the Danes written in Latin in the
Iate twelfth/early thirteenth centuries. Saxo's two descriptions of sharply pointed battle deployments,
assumed to be the sainfylking even if he does not use this or equivalent terminology, have thus gained
172

Agath. Hist. 2.8.8: {v 6i autoiq r1 i6ia tr;s rorqaralan,g olovei ippoAov' 6eAtcotQ yaq iQxu, rai 'ro piv ipnq6o0tov,
6n6oov iq o[u Alyev, oteyav6v te {v rai, nenurcvc.rpivov,rQ navtoOev roie aonLrtntgtneQgal0aq Qairle te &v autouq

173

Agathias' familiarity with tactical literature is explicit at Hist. 2.9.2,'... making the phalanx "at a forward angle", as the
Tacticians might call it' (intxapntov ipngoo0iav, cbg dv oi,rarctrxol ovopaooatev, trlv QaAayya ranaornoag} For other
instances of Agathias' use of technical vocabulary drawn from this genre see e.g. 2.8.4: \;ALoytop6c,, a term otherwise
attested only in Asclep. Tact.2.5; AeL Tact.6.L2.; Arr. Tact.7.7-2.I plan to examine this aspect of Agathias' historical writing
in a separate study.

ouoq rceQaAlv

rf

ouv06oet ar.orvntboao0o.t.

17a See CaurnoN 7970,48-9;ll'NNrano 2004b, 1009-11.


17s Miir-r-sNrHorr 1890-190& IV (1900) 180
176 Krnpr 1901, 349,369 n.7.
177 StillusefulisNrcxrr1915,whoargued(esp.488-95)foradirectlinkbetweensainfylkingandcaputporci(num);expandedin

NrcxEr. 1918,289-349.The most recent review of the evidence, citing additional bibliography, is Bscr 1965, 41-7; summarised
in Brcx 1998, 291.-3. See also Guraxrnuuunn 7954, 373-6.
r78 Drrrnticx 1920-28, II 43-4; Gxossr 1920, 255-6; SeNonn 1939, 8; GuNont 1939, 162; Larr,rrusnr 1940, 29-30; Lanrovusxv 2007, 62;
Wuerrrn 2004 346-9; JaNNreno 2004b,7003-4 ('imprudement rapproch6'). PfnEz Casrno 2001 80-81 does not appreciate the semantic diversity ol cuneus. Attempts at tactical reconstructiory uninformed by philological enquiry or competence, should
be read with caution: e.g. lJNnrnwooo 7999 13L-4; SrnpnnNsoN 2007,31'-6.

76

THE PHILOLOGICAL EVIDENCE FOR GERMANI IN THE LATE ROMAN ARMY

a prominence in wider scholarship out of all proportion to their historical value, especially given their
far-fetched geometrised precision, uncertain sources and legendary character.lTe
In the present author's view, the proposition that the Romans borrowed a specific'wedge-like' deployment from the Germani, and simultaneously coined a Latin translation or rendering of its original
Germanic name, is not consistent with the evidence for earlier Roman and Germanic tactics, even allowing for the possibility of nominal borrowings for broadly similar practices within a shared military
culture (see above for fulcum). Furthermore, Roman military vocabulary abounds in examples of the
metaphorical application of animal names to formations, weaponry or machinery (e.g. testudo, aries, lu-

pus, ericius, onager, ceruuq scorpio, musculus, corrus, murex), and there is nothing intrinsically'un-Roman'
about the coinage caput porci(num) thatwould lead one to suspect a non-Roman semantic antecedent.rso
If caput porci(num) is an expression formulated on the basis of a distant and undocumented terminolog-

ical forerunner of sainfylking in some earlier Germanic language(s), the circumstances of this process
can only be conjectured. The late Roman recruitment of Germanic peoples from across the Rhine as
high-quality infantry might offer a plausible scenario wherein Roman soldiers could have popularly
applied a translated 'barbarian' expression to an existing Roman formation, though this would necessarily assume that a correspondent to (much later-attested) Old Norse soinfylking was found also
among West Germanic-speaking peoples. Unfortunately the extent and the nature of a potential connection must remain obscure, in part owing to insufficient collaboration or dialogue between classical
and Old Norse historical and philological scholarship. In the current state of research, no more can be
said with confidence.

CATEGORY

5:

ERRONEOUS GERMANIC DERIVATIONS

drungus, 6po0yyoq
I have discussed the term drungus (> bqo0yyoq, cf. derivative bqrouyyroti, bqouyyaqrog) at length elsewhere; only an outline of the arguments is offered here.lsl Latin drungas is first attested in the late
fourth century as a synonymfor globus, a generic 'band, crowd, throng', in primarily military contexts,
without specification of whether infantry or cavalry. The loan DqoOlyog is first documented in Greek
in c.404-7 in relation to a 'mob' of enraged monks.182 The term next occurs in Maurice's Strategicon, in
which the usage of bqo0yyoq embraces two principal meanings: first, a non-technical expression for
any 'grouping' of trooPs, either Roman or non-Romary ranging widely in size and type from large

brigade- or division-strength formations of cavalry to a four-man skirmishing party of infantrymen;


second, a specific non-linear cavalry deploymenf characterised as compact and flexible, and suited to
irregular operations, ambushes and surprise attacks, and especially concealed outflanking manoeu-

180
181

782

Saxo Grammaticus, Gesta Danorum 1.8.16 (a fictional battle between Hadingus, an early legendary Danish king, and the
Byarmenses); 7.70.6, cf .8.4.3, 8 (the legendary or at least heavily mythologised Battle of Brdvellir or BrAvalla). See Bacx 1965,
43, 46 lor identification of the ninfylking. It has even been suggested that Saxo's descriptions are in fact Vegetius-inspired
fantasies: Hanrlrz 7939,148-9, although the language in the two works differs significantly.
On this phenomenory with additional exampleg see the intriguing monograph of McCenrNev 1912.

Raucr 2004a.
Latin: Veg. Epit.3.16.3: scire dux debet contru quos drungos, hoc est globos, hostium quos equites oporteat poni ('the general should
know against which drungi, that is groups, of the enemy he ought to position which cavalry'); 19.2: .. . a multitudine hostium
aut a oagantibus globis, quos dicunt drungos, ('... by a mass of the enemy or by mobile groups, which are called drungi'); SHA
Prob. 79.2: omnium gentium drungos usque ad quinquagenos homines ante triumphum da:lt ('[Probus] Ied bands of up to fifty men
from all nations before his triumphal procession'). Greek: J. Chrysostom, ep. 4 ad Olymp.2 (PG 52 614): a0q6ov ilno trlv 6r^r
bqoOyyoq prova(6vtc,.rv - oftr.r yaq 6ei elnelv rcai t1 Ailerrqv pavlav autcov evbelfaoOar - iniotrlorov tr;1 oirciq. tv0a
r;lpev ('collectively towards dawn a "drungus" of monks - for thus must one call it and indicate in speech their frenzy came
to the house we were in'). See discussion in ReNcs 2004a,97-105.

77

PHILIPRANCE

vres on the battlefield. The later Byzantine evolution of bgo0yyoq and cognate vocabulary is of no
immediate concern.rS3
Leaving aside the more outlandish antiquarian etymologies, two competing theses on the derivation of drungus long coexisted. The few historians of the late Roman and Byzantine army who have
discussed or mentioned the term almost universally cite a Germanic derivatiory often in the context
of a perceived'Germanization'of late Roman military practices and personnel. Accordingly, drungus
is associated with the modern German verb dringen (<*pringan), with the sense'to crowd, to pressi assuming a nominal etymon "prunga. The primary inspiration of this thesis is historical circumstance or
convenience: the fourth-century appearance of drungus is thus accounted for by the greater presence
of Germanic peoples within the empire from that date and their presumed influence upon Roman
tactics and fighting techniques.l8a Insofar as any linguistic argumentation is offered, in all cases Du
Cange's lexical entry appears to be the source of this line of reasoning.lss However, even if one wishes
to accept'Germanization' or 'barbarization' as an accurate portrayal of the late Roman army, a derivation of drungus from a posited *prunga is intrinsically unlikely on etymological grounds. The earlier
lexical literature claiming a Germanic root on the basis of superficial phonological similarities with
modern German forms is the product of naive seventeenth-century guesswork, and even the very few
subsequent philological studies to prefer a Germanic derivation have acknowledged its inherent difficulties.186

In contrast, for over a century philologists have concurred with equal unanimity that Late Latin
drungus is undoubtedly of Celtic origin. This is most clearly manifested in Old Irish drong and Old
Breton drogn or drog, whose original sense was 'people' or'tr7be', which came by extension to mean a
'9rouP, muster, troop, crowd, throng, band', both in a generic and a specific military sense. On the basis
of these attested correspondents in Celtic languages, the posited Gaulish "dhrungho is held to be the
etymon of Latin drungus.187 The adoption of a Celtic term in this semantic sphere is entirely in accord
18

References in DTNNIs 1987,523-4, index graec. s.vv. bqoul yLo,r[, bgo01ryoq. Detailed discussion in ReNcr 2004a, 105-20.

'* E.g.OuaN189$175-5;Kurarovsxn79O2,12-14;Gnossr1920,256,'DieseletztereAusdrucklsc.drunguslisteinsdervielen

germanischen Lehnw<irter die damals in die rrimische Soldatensprache eindrangen'; Lanunnr 7940, 28,'eine ... offenbar
germanische Bezeichnung'; HalooN 7984, 385; Kor.r.aurz 1981 108 n. 9, ' Das Wort Drungus ist eindeutig germanischer
Herkunft'; Bartusis 1991, 664,'from the Germanic thrunga (sic)'; Korras 1993,40.
1s Du CeNcr 7678,I11198.
186 The 'Teutonic' origin of drungus was first
asserted by PoNraNus 7606, 223-4, 'Drungus hodieque purum putum nostri
Teutonicique idiomatis vocabulum est . . . Nam hodie nostratibus Germanisque usitatius est, Gedrung & gedreng'. SrsLNaaN
1664,185-6 telated drungus to Anglo-Saxon drunga (in fact unattested) and modern English 'throng' (in fact derived from
Old English geprang).Both Pontanus and Spelman are cited by Du Cer.rcr 1678, III 198, seemingly with neither consent nor
rebuttal, though his comment on Pontanus' reasoning 'quod idem sonat' (i.e. drungus and Gedrung) perhaps hints at caution. Presumably following these entries, vox Mrxr.osrcn 1886, 50 str. drongai, adduces Old English drunga (sic). Subsequent
studies were aware that initial Germanic pr-would produce Latlntr- and not dr- (see Krnpr 1901, 369-70). BnticH 1913,
1,6-17, while acknowledging this impediment, indulged in special pleading to postulate that Latin drungus had therefore
resulted from a conflation of two (unattested) Germanic words "prung and *druht. Moreover, faced with an alternative
Celtic etymology (see below), Briich explained away Old Iri sh drong (and presumably other attested Celtic forms) by boldly
asserting that the Old Irish form was itself, on the contrary, a borrowing from Latiry and not vice versa; for this he cites
Walde, though Walde says nothing of the sort and in fact favoured the Celtic derivation (Warnr-HoruaNx I 374-5; Briich
cites ed.1 [1906] 185). Although at best improbable, Briich's conclusion was cited by Garurrrscurc 1970,23, and his argument
was recently rehearsed by Gnonx 7998 185, though with much less conviction. Ku:olr 7972, 44, noting that the word 'klingt
germanisch', nevertheless recognises that the initial sound is problematic and the Germanic etymology doubtful. Briich
1913 was the last and only modern study to argue the case for a Germanic etymon on the basis of the philological evidence.
In a lengthy review article, Tnnnacrmr 1921,425-6 casually inferred without argument, and purely on the basis of historical
circumstance, thal drungus was likely to be Germanic, because of its late attestation, and suggested (as Briich, implausibly)
that it had thereafter somehow passed via Latin into Insular Celtic languages. Terracini's line of speculation, unscientific
and later emphatically rejected by other philologists, was nevertheless repeated as fact by Wnrscrxsrx 1931, 199; and thence
by Wnarnoucu 1970,894 (posthumously published from an unedited typescript produced in 1950).
187 Holonn 1896,1 7337 ssr. drungo-; Knupr 1901,
369-70; PnornssN 1909-13, | 106-7; TLL V.7 2077.7-2,'vox gallica vel britannica
esse videtur, cf. hibern. drong aremor. drogn"caterva,tutba"'; DoruN 7920,253; Weron-Horrraarlrv I374-5; Poxonrw 1959,
I 1093, s.v. trenk-, 'ngr. bgoOyyog, spdtlat. drungus und air. drong 'Schar' sind echt keltisch (*dhrungho-) und keine germ.';
Mrui.sscu 1968-9,495; EnNour-Mnrllnr 185, 'mot 6tranger, sans doute celtique'; Launrnr 7997, s.v. drong;Ltuwxr 2003,205;
Dueuannr 2003, 126. The Celtic derivation is tacitly accepted by Mrr.urn 7996, 98 n. 4.

78

THE PHILOLOGICAL EVIDENCE FOR GERMANI IN THE LATE ROMAN ARMY

with the wider evidence for the influence of Gallic equestrian practices, technologies and vocabulary
in the historical development of Roman cavalry. Given the fourth-century emergence of drungus in
Latin literature, an undocumented'pre-history' of this loanword in a Vulgar linguistic register can be
assume4 a pattern paralleled in the documentation of other Celtic loans first attested in Late Latin.188
you(iov
The term you(ta. occurs in Maurice's Strategicon in a description of a type of incendiary missile: '... using the so-called 'chouzia' which are shot from stone-throwing artillery and are themselves filled with
fire' (bra tr,-rv Aeyoprdvruv 1ou([o.ru ano netqop6Acov rcai airruoy ftuqoq nerAqqr.rpevrov).18e The context
implies a receptacle filled with inflammable material and shot from an onager, but the precise character
of this projectile is uncertain and it is not clear whether the term denotes a specific device or a generic
category of incendiary'. Maurice is the only author to use this term. When Leo VI reprised this passage
inhis Tactica, he substituted Stones filled with fire using a (flammable) substance'.leo In his annotations
to the editio princeps of the Strategicon in1664, Scheffer posited a Germanic etymology for you(La on the
basis of modern German Schutz.lel Although this linguistic association was at best superficial, a Germanic origin is considered probable by Mihdescu and possible by Kuima.1e2 Formerly a hapax, plural
you(ta is now also documented epigraphically in a list of commodities subject to import duty recorded

on a fifth-/sixth-century inscription from Cilicia. In this context the editors tentatively identified a
type of gourd, probably to be understood as a container of liquids rather than the fresh fruit.1e3 This
discovery requires earlier guesswork to be set aside and a Greek etymology to be entertained, whereby
the projectile was metaphorically named after (rr^rv Aeyopvrov) the gourd or vessel. If so, one might
draw a close historical analogy with the derivation of modern English'grenadei via French'grenadoi
from'pomegranate', owing to a formal resemblance. Of potential relevance is the occurrence in some
metrological texts of the word Xaled or you(a (to my knowledge unregistered in modern lexica) as a
gloss for Xoig, an ancient and well-documented liquid measure, although the currency of lo0q itself
does not appear to extend beyond the Ptolemaic era and a direct connection cannot be demonstrated.lea
In any case, the alleged Germanic etymology is demonstrably false and there is no need to consider
further the nature of this incendiary device or its alleged (and historically unlikely) origin in Germanic
siegecraft.les

sculca, orcoOlxa

I have dealt with the etymology of this term in detail elsewhere and give here only an outline of the
arguments and most important studies.le6 A species of Late Latin termini technici, some directly attested
in Latin sources, others witnessed only as transliterations or loans into Greek, first emerges in the early
sixth century in connection with intelligence gathering and defensive surveillance. Nominal sculcal
orco0Arca variously signifies the activity of 'reconnaissance', a 'scouting party' or 'patrol-duty', with
188 See Rexcr 2004a, 100-105 withbibliography. See also Kurapr 1901, 348,367-8;TLL117797.70-74,s.:v.bebra.
18e Maur. Strat. 10.1.514.
1e0 Leo, Problemata 10.6 (Dain) is simply an excerpt of Strategicon 10.1, retaining
1ou([<.rv. Leo, Tactica 15.26 (Dennis) substitutes
tdrv netg6v nuqoq rrenArlqrc.rpfvov b16A1g. Cf. also Tactica79.63: earthen pots (1litqag) filed with a flammable preparation

in naval combat.
1664, 480-81,'videntur 1ou(ia haec convenire cum schutz Germanorum meorum . . .'.
1e2 Mrn,i.rscu 1968-9,1493;1970,25L; thence Kudue 2004,175 n. 1. An alternative derivation proposed by Du CeNca 7688,1758
frornl.atin excusslo need not detain.
1e3 Dacnou-Fercset 1987, U1, Nc 108.6, wlth \77; now also LSf rev. suppl. 3\4, s.v.1oil(Lov, 'a kind of gourd?'.
tea Cf. Epiphanius, De mensuris etponderibus (de Lagarde 1223.22 = Sakkelion 131.34): Xong iortyovla, to piv pya eot6v
eirooL rai orto; Etymologicum Gudianum (Sturz), s.v. 1qo0q (slc); cf. Heron, De mensuris 60.4 (Heiberg). For historical usage

rer Scnerrsn

of louq see F{ur-rscs 1882,101,104,107-8, 624-9; Scnttsrcrtl970, 114-15.

res See RaNcn (forthcoming) for further discussion.


re6 RaNcs 2014.

79

PHILIPRANCE

cognate oKouAKilew 'to reconnoitre', and orouLrcorr,uilq, 'a scout', as well as adjectival sculcatorius and
orcouArcatoq, and compound forms ngoorcouArcefr^; and nqroorcouArcritoqr. Scholarship long acknowledged an etymological relationship with Late Latin exculcator, documented as a generic'skirmisher' in
the late fourth and early fifth centuries, but largely overlooked the significance of proculcator, 'scout'
or'guide', attested from the mid third century.leT From the early twentieth century, some philologists
and more lexicographers endorsed a Germanic or specifically Gothic derivation. This view was first
delineated en passant in a review article by Jud (1908), and later elaborated by Meyer-Liibke (1935) and
Gamillscheg (1934-6), who posited a Germanic root *skulk- on the basis of alleged correspondents in

medieval and modern Scandinavian languages.le8 Their indices for a Germanic or Gothic derivation
were widely cited in Latin and Romance etymological lexica without further enquiry into the complex
evidence and arguments, though the lexicographical literature frequently exhibits obvious confusion
and inconsistency.lee This influential thesis, often rehearsed uncritically, has sustained preconceptions
about the'Germanization' of the late Roman army and, more specifically, contributed to studies of the
philological impact of the Goths and/or Lombards, particularly on the language and toponymy of ear$
medieval Italy.200 The alleged Germanic derivation continues to be endorsed in more recent studies,
despite its intrinsic tenuity and even though it was shown to be entirely misconceived more than three
decades ago.'o'
At a historical level, sculca and its cognates appear to encompass typical and long-standing Roman reconnaissance and security procedures, while no Germanic specialism or even capability in this
sphere could be identified. Accordingly, Maurice, although in some measure indulging in ethnographic
stereotyping, characterises contemporary Germani as personally brave but tactically unsophisticated
warriors, who, of all the Romans' opponents, most easily fall victim to ruses, ambushes and surprise
attacks precisely'because they give no consideration at all to reconnaissance (sculca) or other precautions' (... pq ndvu Qqrovt[(ovteq orcotiArcaq rcai tr;e Lotnqe aoQaLeiag).2o2 !'\ire are therefore required
to accept a Germanic derivation of a Late Latin term for practices that are characteristically Roman but
explicitly un-Germanic. Even if this anomaly could be set aside, however, the linguistic arguments do
not survive scrutiny. Little-known comparative studies of the Germanic and Romance evidence by
Betz (1980) and Meier (1981), not cited in and apparently unknown to subsequent scholarship, exposed
the Germanic etymology as baseless.203 The problem originates with Jud, who first posited a Germanic
verb "skulkan on the basis of Middle English skulken and modern Swedish skolka, which in his view
meant '6pier,6tre ir l'aguet'. Meyer-Liibke also cites Middle English skulken as well as Old Norse skolka
(in fact unattested), which he claimed meant'sp?ihen. Gamillscheg went further in positing nominal
Gothic "skulka, purportedly 'SpZiheri on the basis of two verbs, modern Norwegian skulka, which he
understood to mean'lauerrl finster blickeni and modern English skulk, similarly construed as'auf der
Lauer liegen'. Firsf always problematic for this thesis was its exclusive reliance on North Germanic
forms (Scandinavian languages) and loans therefrom (Middle English skulken is a later borrowing from
Old Norse) in contrast to a complete lack of parallels in West Germanic languages or, more tellingly,
Gothic. The medieval and modern Scandinavian (North Germanic) forms cited in support of a Gothic
1e7 Al1 references and discussion in RaNcs 2014.
"u Juo 7908, 460-6L; Mavcn-Lrisxs 1935, 640 $7753a; Gerr,rrrrscnrc 1970 [ed.1 7934-61,1392,1154; reiterated in GaurrrscHrc 1939,
92.
1ee The bibliography is vast; I mention only the more
often-cited works: Zttttecus 1935, 160-67,'... sculca, das vielleicht nordgermanischen Ursprungs ist', also 234; War.on-HoFMANN II 502 s.v. sculca, '... wohl Lehnwort aus Germanischen (mengl.

skulkery an. skolka "ausspiihen")', in apparent preference to an alternative Latin derivation; MrcrronrNr-Duno 1964, s.v.
ll 369ab; Enr*our-MsrI,I,nr 606, s.v. sculca; Coxov.rlas-Pascuar 1980-83, II
778-719. See further the largely derivative lexicographical literature assembled by Mnrnn 1981, 155 (Italian), 157 (Spanish).
200 Fot the Italian place names Scolca, Sculcola,
Scruccola: e.g. SananrNr 7963-4,141-744; Brnroulr 1968,494-6; Masrnnru 1973,
647-53; Rrsrtr.r.r1984,152-62;Casrnr.r.aNr 200Q 42-3; Monuccuro 2003, 158-60. For referencesto sculca in Lombard law codes:
e.g. vAN oen Ruar 1970,119-L20.
201 See recently e.g. Bnronn 2000, 226; cited,
with approval by Drrruanr 2006, 24.
202 Maur. Strat. 77.3.30-32; cf. similarly 4.1,.15-77.
n3 Bntz 1980; Meier 1981.
scolta (sic),'dal germ. *skulka"; Conouruas 1954,

80

THE PHILOLOGICAL EVIDENCE FOR GERMANI IN THE LATE ROMAN ARMY

(East Germanic) antecedent share the common feature of an intensifying suffix -k (skul + k), whose
consequent resemblance to sculcaforcouArca provides the main rationale for inferring a Germanic etymology. Betz argues persuasively that this morphological development was both exclusive to the languages of Scandinavia and postdates Late Antiquity. This accounts for the dearth of correspondents
in all other Germanic linguistic sub-groups, except for a medieval loan into English, but also rules out
a link between these Scandinavian forms and Late Latin sculca and its cognates. Second, the medieval
and modern Scandinavian forms adduced in support of this etymology simply do not have the meanings that older scholarship wished to assign to them. In fact, none of the cited forms has any connection with'scouts' or'spying', or being'on the look out' or'lying in wait'. These witnesses to the radical
*skulk- in Scandinavian languages do convey a sense of hiding or evading notice, but with a different
and primarily negative connotation, that is to say, with the intention of avoiding a duty or obligatiory

skiving or shirking, sneaking off or playing truant. Accordingly, the proposition that posited Gothic
l"skulka was the etymon for Latin sculcarests on nothing more than misconceptions, bold conjecture and wishful thinking.2oa
Furthermore, proponents of a Germanic etymology always struggled to explain or accommodate
exculcator (undeniably cognate with orcouArcarrog and thus with orco0Arca), which clearly derives from
Latin exculcare (ex + calco),'to tread ouf to trample'. Still more problematic is the existence of directly
analogous proculcato1 from proculcare (pro + calco),'to tread forward', in the sense of a guide or advance
party 'beating a path'. More to the poinf proculcfitor, the earliest-attested representative of this lexical species, shows that the common root is in fact culc- and that the search for an etymon of *sculc-,
Germanic or otherwise, is entirely misconceived. In reality, a well-documented phonological shift in
Vulgar Latin accounts for exculcator >*sculcator > orcouArcatr,-rqr, for which contemporary Latin military
terminology offers direct parallels, notably excubitor > scubitor > orcoup[.rr^r9.205 Correspondingly, Late
Latin sculca derives unproblematically as a deverbal noun frorn exculcare > *sculcare > sculcaf orcoiLrca.2o6
Thecognateverballoanalsofollowsawell-attestedpattern: exculcare>*sculcare>orcouArcedeLv.207Within the Latin-based argot of the East Roman army a close parallel may be drawn with the deverbal noun
roga, a technical term for soldiers' pay, derived frorn erogare > rogare > rogalQ6ya, with a corresponding
verbal development erogare > rogare > QoyerieLv.'z08 It should now be clear that there was never, in fact,
either room or need for a posited Germanic root*skulk-, which has, in any case, been shown to have no
foundation.
*skulkan

CONCLUSION
The foregoing survey has identified a small number of Germanic loanwords in Roman military vocabulary. Some are borrowed names for borrowed things: tufa, carrago, armilausa. Others appear to be
nominal borrowings, notably fulcum, or cases in which the specificity and motivation of the loan are
unclear: bandum, pungalnouyylov, QAaorclov. The case of framea should be set apart on the grounds
that this loan is a purely literary development occurring at a much early date and unrelated to the
actualities of Roman-Germanic interaction in military culture. Of the cases of uncertain or doubtful
Germanic etymology, insofar as the evidence permits conclusions, I am inclined to admit only barritus,
while recognising the difficulties of all scholarly positionq past and presenf on this question. I reserve
judgement on to0Aboq, though on currently available evidence a Latin etymon seems more plausible.
2oa 8nr21980,49-51,. See also Msrsn 1981, 161n.7, citing pers. comm. with Belz,'Germ.*skulk'Spdher'ist also zu streichery und
es bleibt nur noch erkliiren, wie fud so viel Glauben dafiir finden konnte'.
205 References and detailed argument in RaNcs 2014.
206 This was, in essence, the original deduction of Du CaNcs 1678, lI 410-77, ssr. collocare; YlL, p. 375a, s.v. sculcae. See also Du
CaNce 1588, 7396-7, s:r. orodArcaq with modifications by Mruixscu 7978-9,111368-9, see also I 212.
207 See the numerous examples collected by PsarrEs 1913,31.6-18.
208 Knarrann 1992.

81

PHILIPRANCE

A Greek etymology of burgus (< nriqyoq) ig in my view, less problematic than a Germanic derivatiory
while hornatores is most plausibly explained as transcriptional error. In the cases of posited *cautia
and the phrase caput porci(num), the arguments cannot extend beyond tenuous speculation. The final
category requires no further comment beyond reiteration that drungus, you(Lot and sculca are demonstrably not Germanic loanwords.
These loanwords differ somewhat in character and application. Some were termini technici whose
currency does not appear to have extended far, if at all, beyond an institutional usage within the
affnyi carrago, armilausa, fulcum (and this is true also of 'ro0A6oq, whatever its derivation). If correctly
construed, barritus, in the narrow sense of a Germanic war cry, seems to belong to the same milieu,
although the form found in late authors would be the result of a popular etymology or association
(barditus > barritus) in sermo castrensis. Other loanwords appear to have had a more pervasive impact:
derivatives of bandum and tufa are attested in Romance languages to a degree that suggests a broader
penetration in Vulgar Latiry at least regionally,2oe while it seems unlikely that pungalnouyylov and
flascalflascolQlaorclov ever existed in exclusively or even predominantly military contexts.
Bearing in mind preceding remarks about the time lag in the documentation of loanwords, a chronology can be reconstructed only with caution. It is notable that framea alone, already distinctive for
other reasons, is directly attested in Latin before the late fourth century. In contrast tufa, armilausa
and carrago all first emerge towards the end of that century, in respectively Vegetius (writing 383-450,
in my view 390s), Paulinus (398-9) and Ammianus (c.390-91). Similarly, apart from Tacitus' reference
to barditus as a Germanic custom in his ethnographical treatise, the earliest evidence for barritus as a
practice and alleged Germanic loanword in the Roman army is also found in Ammianus. Within this
chronological grouping, the evidence for carrago differs slightly inasmuch late fourth-century authors
merely apply this Gothic word to a Gothic custom; it is not until the later sixth century that a Roman
author applies this term (as rcaqa.y6q) to a Roman practice, although the historical evidence indicates
that the East Roman army had adopted this procedure by the early sixttr, and possibly late fiftl1 century. Other termsfulcum,bandum,pungalrouyylov and QAaorclov are not documented until the second
half of the sixth century, but of course this by no means excludes the possibility, even likelihood, of a
linguistic loan at an earlier date.
The evidence also permits a degree of geographical distinction. The documentation and historical
contexts of barditus-bqrritus point to western Germani, recruited as infantry into the Roman army,
principally as auxilia palatina from the late third or early fourth century onwards. For reasons outlined
above, this is also the most plausible scenario for the loanword fulcum. The linguistic evidence for
armilausa and tufa does not permit regional or cultural differentiatiory even if an East Germanic (or
specifically Gothic) cultural context for tufa seems more attractive on historical grounds. The Latin
and Greek cognates of QAaorclov allow for a multi-stage loan from both West and East Germanic. In
contrast, three loanwords - carrago, bandum, pungalnouyylov - explicitly or demonstrably derive from
a Gothic etymory and appear to relate to specific practices or features in the equestrian and military
culture of the Goths. The chronology and distribution of these loans suggest that they were adopted
into the Latin-based argot of the East Roman army in the Balkanq at some point between the early fifth
and mid sixth centuries. Subsequently, it seems thatbandum andpunga were transferred to the West,
notably to the Italian peninsula, by the protracted campaigns of Justinianic armies, a pattern that is
replicated in the documentation of other items of East Roman military jargon, even those that arose
from organic developments within Latin (e.g. sculca, roga).210
The most likely environment for linguistic loans in this semantic sphere is the more frequent
employment of Germanic warriors by the Roman government in different circumstances and under
various terms of service - regular soldiers, pederati, bucellarii or allies - from the late fourth century
Bnticu 1913, 19,58,118,139,162;Mwtt-Liisxr7935,76
Gar,rrr.r.scurc 197 0, Il 257 n. 2; Monr.rccuro 2002.
See

82

Krar,rrn 1992; ReNce 2014.

59929,930;7a7 58973;

Mrrrirscu 1968-9,1488-9;7978,32,305 ($$22,309;

THE PHILOLOGICAL EVIDENCE FOR GERMANI IN THE LATE ROMAN ARMY

onwards. The transfer or exchange of nominal loans at this level does nof of course, imply or require
bilingualism nor indeed any depth of cultural-linguistic understanding or insight. The small number
of Germanic loanwords should not surprise, both in general and in specific military contexts. The
greater prestigg sophistication and complexity of Roman culture meant that linguistic loans generally flowed towards Germania: around 400 Latin loanwords, predominantly from Vulgar Latin, have
been identified in Germanic languageq in contrast to around a dozen Germanic loanwords directly
attested in Latin texts, mostly in relation to trade or warfare. These may be supplemented by up to 100
or so Germanic loans into Vulgar Latin pre-400 AD posited on the basis of correspondents in Romance,
although in many cases such an early chronological horizon is far from secure. Even allowing for the
fact that some Latin loans into Germanic languages are likely to be due to the later influence of Christianity, the dominant direction of loanword traffic is clear.211 Within the framework of military institutions, historically there is a tendency for'foreign legionsi broadly construed, to acquire a functional
proficiency inthe Heeressprache of their imperial paymasters, rather than increase the currency of their
native vocabulary. The overall superiority of the Roman army in equipmenf logistics and organization would also militate against Germanic influences, both technological and terminological. For the
sake of comparisory one might observed that around twice as many loanwords from Celtic languages
are identifiable in Latin military terminology, which potentially reflect a longer-term and more profound cultural engagement.2l2 The implications of such calculations defy easy assessmenf however,
and Roman contacts with other peoples caution against drawing simple conclusions on the basis of
the linguistic evidence alone. For example, the significant influence of nomadic peoples of the Steppe
- Huns, Bulgars, Avars - on Roman bows, archery kit, and equestrian equipment and techniques is
widely acknowledged and, in part, demonstrable archaeologically, but a concomitant linguistic debt to
Altaic languages in these spheres is much harder to trace.213 Similarly, the East Roman army could, for
instance, borrow explicitly 'Slavic' or 'Moorish' designs of javeliry without also adopting their respective Slavic or Berber terminology.2la
Only two of the Germanic loanwords relate to operational proceduresi carrngo, a novel, culturally
specific Gothic practice copied by the Romans, and fulcum, a loanword which, on the contrary, appears to have been motivated by broad correspondence in Roman and Germanic tactical deploymenf
even if the loan may entail a degree of misconception. All the other Germanic loanwords belong to
the sphere of "military culture" - dress, personal kit, emblems, ceremony and ethos. Here historical
antecedents supply broader contexts. Both tufa andbandum, for example, follow a Roman tradition of
appropriating new or exotic designs of standard or crest from foreign peoples, most notably the draco,
probably imitated from the Sarmatians or Dacians during or after Trajan's campaigns in the early
second century, although it is notable that, in contrasf this'dragon-headed' standard was accorded a
Latin designatiory while its original terminology, in whatever language, was discarded and remains
unknown.2rs Correspondingly, in this domain there was a terminological counterflow from the Ro2rr Latin loanwords in Germanic languages:

see selectively Fnrr.rcs 1966-8; Wtro 1976; GnsuN 199& esp. 201ff. Specifically Latin
loans in the military sphere: Gamillscheg 1970, 18; KunN "1972, esp.14-19; GnrrN 1998,232-5. Surveys of Germanic loanwords
in Latin: Bniicn 1913; GnEEU 199$ esp. 183-200
212 Krupp 7907,367-8,385-6; RaNcn 2004a, 100-105.
2r3 To my knowledge, only one possible Altaic linguistic loan has been identified in relation to Roman archery and equestrianism: Maurice (Strat. 1.2.1.6;12.8.5.2.4) applies the term rcorircouqov to a quiver (cf. German Kocher, Ftench carquois).By this
date Roman cavalry used a quiver originating in Central Asia, in the form of an elongated box with a hinged lid, usually
waisted above the centre and broadening towards the bottom to accommodate the fletchings of the arrows, which were
drawn from the mouth point-first. This design was probably introduced by the Huns from the 370s, though the Avars may
have inspired modifications from c.560s: see Coulston 7985,277-5; Kolias 198$ 228-9;Dwyer 199& 81-8. The etymology of
rcoJrcouqrov has been indentified as Hunnic or Avaric: Hubschmid 1953; Maenchen-Helfen 7973,427; BaNer 1986, 1.5, contra
a Germanic origin favoured by TnreNoanHyr,I.rDrs 7909, 770; Mrniescu 7978-9,11 59.
214 Maur. Strat.12.8.5.6 (Aayrcibta lrAaprviorcta), 20.10 (arcovtiorq pLrcgoig MauqLorclorq).
21s See Cour.sroN 1991 with older bibliography. Another example might be the so-called cantabrum, a type of standard poorly
documented in the late second/early third centuries, whose etymology suggests a Spanish origin; for references: TLL III
278.68-72.

83

PHILIPRANCE

man world into barbaricntn, as evidenced by the reception of Latin loanwords in Germanic languages:
Latin draco > Old High German trahho, Old English draca;Latin signum > Old English segn.216 As previously observed, the Central Asian parallels to the tufa illustrate the possibilities for the long-distance
cultural exchange of symbols and emblems of military prestige. Similarly, the use of elaborate harness
ornamentation by Roman cavalry was an ancient practice, but new decorative styles, like the animalhair tassels also encompassed by the term tufa, were doubtless assimilated in imitation of flamboyant
barbarian warriors.2l7 Late Roman adoption of a Germanic term for a Germanic item of clothing such as
the armilausa also finds parallels in earlier Latin military vocabulary, notably sagum, from a Celtic roof
which became a basic, even defining, item of a Roman soldier's dress. Consideration of the longue durie
of Roman military eclecticism underlines the meagre number of Germanic loanwords in this sphere,
but their character reinforces the impression gained from the wider evidence for Roman-Germanic
interaction that, although at a technological or operational level there was little or nothing West or East
Germanic peoples could teach the Romans about the Art of War, they nevertheless supplied the empire
with sources of high-quality professional warriors and the sights, sounds and fashions of their martial
traditions became ingredients of the institutional Mischkultur of the Late Roman army.

REFERENCES
Arporpr 1950

Arrornr, A.: Etudes sur le Tr6sor de Nagyszentmikl6s. Cahiers

Arporpt

Arch6ologiques. Fin de lAntiquit6 et Moyen Age 5 (1950) L23-49.


Arporpr, A.: Cornuti: A Teutonic Contingent in the Service of Constantine the
Great and Its Decisive Role in the Battle at the Milvian Bridge. DOP 13 (1959)

1959

Avronv 7996

Baarz

1981

Barpwl.r 1977
Baur.rr 1978

L69-83.
Avronv, P.: People and Identity in Ostrogothic Italy 489-554. Cambridge L996.
Ba.xrz, D.: Burgus. RGA2 IV (1981) 274-6 (repr. rn idem, Bauten und Katapulte
des romischen Heeres [Mavors XI]. (Stuttgart1994) 83-5).
BarowtN, B.: Theophylact's Knowledge of Latin.Byzantion 47 (1977) 357-60.
BAuNt, C.: Vestiges arch6ologiques de 1'6poque tardive des Sassanides et leur
relations avec les peuples des Steppe. Acta Arch. Acad. Scien. Hungarical 30
(1e78) 173-212.

BaNpr 1986

BanroN 1985

Banrusrc

L991,

Banpry 1980

BBcr 7965

Brcx 1976
Becr 1998
BsnrorrNr 1968

BaNrr, E.: Problemi di lessico balcanico. Di alcune contrnuazioni del lessico


militare nel neogreco e nelle lingue balcaniche. In: Studi Albanologici, Balcanici, Bizantini e Orientali in onore di Giuseppe Valentini, S.J. Studi Albanesi.
Studi e Testi 6. (Florence 1986) 7-29.
BanroN, P: On Making a Roman Cornu. In: Roman Military Equipment: The
Accoutrements of War. Proceedings of the Third ROMEC Semirtar, Nottingham 1985. Ed. M. Dawson (BAR int. ser.336). (Oxford 1987) 29-39.
Banrusrs, M.C.: Droungos. In Kazdhan, A. (Ed.): Oxford Dictionary of Byzantium. (Oxford 1991) 1664.
Barurv, I.M.: The Old English Translation of Orosius (Early English Text Soci-

ety 6). (Oxford 1980).


Bucx, H.: Das Ebersignum im Germanischen. Ein Beitrag zrtr germanischen
Tier-Symbolik (Quellen und Forschungen zLLr Sprach-und Kulturgeschichte
der germanischen Volker N.r. 1,6 - 140). (Berlin 1965).
Bncx, FI.: Barditus. RGA2 II (New York-Berlin 1976) 52-3.
Brcr, H.: Fylking. RG,# X (New York-Berlin 1998) 291-3.
Bsnrorrxr, O.: Ordinamenti militari e strutture sociali dei Longobardi. In: Italia in Ordinamenti militari in Occidente nellAlto Medioevo Settimane di Studio del Centro Italiano di Studi sullAlto Medioevo 15. (Spoleto 7968) 429-607.

216 Scsnaur'r 1954,1247-51.;KuttN 7972,76, 18-L9,46; Gnrrr.r 1998,233-4 with remarks on alternative contexts at283-4.
217 For earlier practices: Sournnnx-Drxox 1992,34-5; Brsnop-Cour.sroN 2006, 120-21,,190.

84

THE PHILOLOGICAL EVIDENCE FOR GERMANI IN THE LATE ROMAN ARMY

Bs5svrrnv

Brrz

1964

BrSrvrrnv, V.: Spdtgriechische und sphtlateinische Inschriften aus Bulgarien.

(Berlin 1964).
Brrz, W.: Romania Germanica atque Romana: Gab es einen gotischen skulka,
Spiiher? In: Romanica Europaea et Americana. Festschrift fiir Harri Meier 8.
Januar 1980. Ed. H.D. Bork, A. Greive, D. Woll. (Bonn 1980) 48-54.
Brcor, L.E. (ed.): Paladii ... de Vita S. Johannis Chrysostomi Dialogus. (Paris

1980

Brcor

1580

1580).

BrNnsn

2000

Brnorn, V.: Sprachkontakt und Diglossie. Lateinische Wtirter im Griechischen


als Quellen fiir die lateinische Sprachgeschichte und das Vulgdrlatein. (Ham-

BtsHop-Coursrorl
DnN Bosrr et al.

1991

2007

BnslrNaN

B6u,t

2006

1980

Boswot'ru-lToLttp,1972
BnowN

1984

Bniicr

1913

Bnucxr.rnn

BnrNuaN, P.: Zosimos II.34.1 and "The Constantinian Reform": Using Johannes Lydos to Expose an Insidious Fabrication. In: Lewin A.S., Pellegrini, P.
(Ed.): The Late Roman Army in the Near East from Diocletian to the Arab
Conquest. BAR int. ser. 1717. (Oxford 2007) 2ll-18.
B6N,o, I.: Studien zum friihawarischen Reitergrab von Szgev6r. Acta Arch.
Acad. Scien. Hungarical 32 (1980) 31-95.
Boswonrrr, J.- Tolrrn, T.N.: An Anglo-Saxon Dictionary, rev. A. Campbell. Ed.a

(Oxfofi1972).

1964

scher Philologen und Schulmdnner in Basel im Jahre 1907. (Basel 1907) 65:77.

Clurnory A.- Cenrnory A.: Christianity and Tradition in the Historiography


of the late Empire. CQ Ns 14 (1964) 316-28.
C-tlrunory A.: Agathias. (Oxford 1970).
CastarraNr, A.: Grammatica storica della lingua italiana I. Introduzione. (Bo-

1970
CasrsrreNr 2000
CarurnoN

1972

logna 2000).
Cnslunory I. (Ed): Strabonis rerum geographicarum libri XVII. Geneva 1587.
Cuanlrs, M.B.: Vegetius in Context. Establishing the Date of the Epitoma Rei
Militaris Historia Einzelschriften 194. (Stuttgart 2007.)
Creusory G.: An Etymological Dictionary of Pre:Thirteenth-Century Turkish.

1957

Crresrl

1587
CHenrss 2007
C-lsAusoN

CrausoN

BrsHor, M.C.- Coulsrory |.C.N.: Roman Military Equipment from the Punic
Wars to the Fall of Rome. Ed.'z(London 2006).
onx Bonrl J.- rrrrr HrNcsl, D.- Trrrran, H.C.: Philological and Historical Commentary on Ammianus Marcellinus XXI. (Groningen 1991).

Bnowry T.S.: Gentlemen and Officers. Imperial Administration and Aristocratic Power in Byzantine Italy A.D. 554-800. (Rome 1984).
Bnticn, J.: Der Einfluss der germanischen Sprachen auf das Vulgdrlatein. (Heidelberg 1913.)
BnucxNrn, W.: Uber den Barditus. In: Festschrift zur 49. Versammlung Deut-

1907

Ceunxox-CaunnoN

burg 2000).

(Oxford1972).
Crnesnv

CorrrNsr

1930

Cororrano

2008

ConorrIN,ts

R.: An Icelandic-English Dictionary, rev. G. Vigfusson, W.A. Craigie.


Ed.'z(Oxford 1957).
CorrrNrl, P.: Sur l'expression OI EN TOII TOT AOII, AIIEPXOMENOI,'teux
qui partent dans les bagages". In: M6langes Charles Diehl. Paris 1930,149-54.
Corouro, M.: Constantinus rerum nouator: dal comitatus dioclezianeo ai palatini di Valentiniano I. Klio 90 Q008) 124-6L.
Conorrrues, J.: Diccionario critico etimol6gico de la lengua castellana. (Berne

1954

1954) (repr.1970).

Conourues-Pr.scuar
CoursroN

1980-83

1985

1991
D,q,cnoN-FtrssaLt987
Coursroru

Conourues, ].-Pe.scuar, |.A.: Diccionario critico etimol6gico castellano e


hisp6nico. (Madrid 1980-83).
Coursrory j.C.N.: Roman Archery Equipment. In: Bishop M.C. (Ed.): The Production and Distribution of Roman Military Equipment. BAR int. ser.275.
(Oxford 1985)220-366.
Coursrory J.C.N.: The Draco Standard. JRMES 2 (1991) l0l-14.
D-c.cnoN. G.-Frrsstr, D.: Inscriptions de Cilicie. Travaux et M6moires Monographies 4. (Paris 1987).

85

PHILIPRANCE

D,c.rl.r

2003

Dn.rrr, F.: Avars and Avar Archaeology. An Introduction. In: Goetz, H.-W.].- Pohl, W. (Ed.): Regna and Gentes. The Relationship between Late
Antique and Early Medieval Peoples and Kingdoms in the Transformation of
the Roman World. (Leiden 2003) 463-570.

Jarnut,

DarN1950
1920-28

Dsr-rnii,cx

Dnreuennn

1993

Drxr.rrs

1981

DrNNrs

1982
1984

DrrrrNrs

schichte. Ed.3 (Berlin 1920-28).

2003

Dnrnsv

Drrsrrlrrncrn

Darry A.: "Touldos" et "Touldon'dans les trait6s militaires. In: M6langes


Henri Gr6goire. (Brussels 1950) [ 161-169.
Dunniicr,; H.: Geschichte der Kriegskunst im Rahmen der politischen GeDrreuannr,

2001

X.: Dictionnaire de la langue gauloise: une approche linguistique


du vieux-celtique continental. Ed.2 (Paris 2003).
Dnrrtv, N.: Apollinaris Sidonius, Carm. 22: Burgus Pontii Leontii: Einleitung,
Text und Kommentar. (Berlin-New York 1993).
DruNrs, G.T. (ed.): Das Strategikon des Maurikios. Ger. trans. E. Gamillscheg
(Corpus Fontium Historiae Byzanticnae 17). (Vienna 1981).
DrNNrs, G.T.: Byzantine Battle Flags. BF 8 (1982) 51-9.
DrNNrs, G.T.: Maurice's Strategikon: Handbook of Byzantine Military Strategy. Eng. trans. (Philadelphia 1984).
DrrssNssncnn, M.: Topographie und Gemeinschaft in der Vita Seoerini.In: Eu-

gippius und Severin. Der Autor, der text und der Heilige. Ed. W. Pohl, M.

DrrrHlnr

2006

Dtuz1887
voN Dorvrlszrwsrr

1885

1920

Dorurt

DusurssoN-Scnanrp

Du Carvcn

1678

Du

1688

C,c.Ncn

2005

1998
1993

Dwvsn

EsnrsruaxN

1996
Enuour-Murrrrr
ErroN

1910
1955-8

Frrucnn
Fnrr.rcs

Furer-Honrucsr

1969-84

Geurrlscnrc

1939

GeurrrscHrc

1970

Grnv ef.

a|.1943-55

1996
GnssN 1998
Gornswonrnv

Diesenberger. (Vienna 2001) 77-97.


DurHanr, ].: Zu neutralen Abstrakta auf -dtov im byzantinischen Griechisch.
JOB s6 (2006)13-26.
Drcz, F.: Etymologisches Wiirterbuch der romanischen Sprachen. Ed.s Bonn
1887 (repr. Hildesheim 1969).
vor.r Dolreszrwsxr, A.: Die Fahnen in rcimischen Heere. Abhandl. des arch.epigr. Seminares der Universitdt Wien 5. Vienna 1885 [repr. In: Aufsdtze zur
rtimischen Heeresgeschichte. Darmst adt t972, L-80).
Dorury G.: La Langue Gauloise. Grammaire, Textes et Glossaire. (Paris 1920.
[repr. Geneva 1985]).
Dunurssory M.- ScHaur,, j. (Ed.): Jean Ie Lydien" Des magistratures de lttat
romain. (Paris 2006).
ou FnrsNr ou CnNGr, C.: Glossarium ad scriptores mediae et infimae Latinitatis. Paris 1578, rev. ed. L. Favre. (Niort 1882-7.)
ou FnrsNr ou Caucr, C.: Glossarium ad scriptores mediae et infimae Graecitatis. (Lyons 1688 [repr. Bratislava 1891;Graz1958)).
Dwvrn, B.: The Closed Quiver. ISAA 41 (1998) 81-8.
EHnrsulury O.: Volk. Mediiivistische Studien zur Semantik und Pragmatik
von Kollektiven. (Gcippingen 1993).
Errory H.: Warfare in Roman Europe AD 350-425. (Oxford 1996).
EnNou,r A.- Mrrrrrl, A.: Dictionnaire ftymologique de la langue latine, aug.
J. Andr6. Ed.a (Paris 1979).
Fmnrcrn, O.: Framea. PW VII.1 (1910) 81-2.
FnrNcs, T.: Germania Romana. Ed.'z G. Miiller (Mitteldeutsche Studien 19.1-2).

(Halle 1965-8).
Furer, S.-HonrucHr, K.: Taq-i Bustan. The Tokyo University Iraq-Iran Archaeological Expedition Report 10,13,19,20. (Tokyo 1969-84).
GenrrrscHrc, E.: Zur Geschichte der germanischen Lehnwiirter im ltalienischen. Zeitschrift fiir Volkskunde 10 (48) (1939) 89-129.
Ge*rrrrscnec, E.: Romania Germanica: Sprach- und Siedlungsgeschichte der
Germanen auf dem Boden des alten R6merreiches. Ed.'z(Berlin 1970).
Grnv, A- Pnou, M.lTussrnn, G.: Recueil des actes de Charles II le Chauve, roi de
France. (Paris 1943-55).
Gor-oswonruv, A.K.: The Roman Army at War 100 BC-AD 200. Oxford 1995.
Gnrrry D.H.: Language and History in the early Germanic World. (Cambridge
1ee8).

86

THE PHILOLOGICAL EVIDENCE FOR GERMANI IN THE LATE ROMAN ARMY

1920

Gnossn

GuNnrr

1937

GuNoEr

1939
1954

GurrNsnuNNnn

vox HerNrscu

Militdrgeschichte von Gallienus bis zum Beginn der byzantinischen Themenverfassung. Berlin 1920 (repr. New York 1.975).
GuNorr, H.G.: Untersuchungen zur Taktik und Strategie der Germanen nach
den antiken Quellen. Diss. Univ. (Marburg 1937).
GuNorr, H.G.: Der Keil in der germanischen Feldschlacht. Gymnasium 50
Gnossn, R.: Rdmische

1939

HerooN

1984

HernoN

2000

Hersau

2001

(1e3e) 154-65.
GurnNnnuNrvsn. S.: Heldenleben und Heldendichtung. Eine Nachlese zu Otto
Hoflers Rrikbuch. ZfdPhT3 (1954) 365-406.
von HesNrscn, E.: Wcirterbuch zu Manghol un Niuca Tobca'an (Yiian-Ch'ao

Pi-shi) Geheime Geschichte der Mongolen. (Leipzig 1939).


Harrory J.F.: Byzantine Praetorians: An Administrative, Institutional and Social Survey of the Opsikion and Tagmata, c. 580-900. (Bonn 1984).
Hr.r.oory J.F.: Theory and Practice in Tenth-Century Military Administration
Chapters 1I,44 and 45 of the Book of Ceremonies, Text and Translation. TM L3
(2000)20t-352.
H-a.rse.n, G.: Warfare and Society in the Barbarian West, 450-900. (London
2001).

1939
Hsnanus 1902

Hlnsrrz

Hsnorn

Hanurz,

1941
1969-70

HorrueNN

1896
1663
HoNy 1984
Hororn

Horsrr

Hurscnuro

1953

Hulrscr 1882
Inu 1897
Is.c,A.c

1992

G.P.: Svinfylkingen. Maal og Minne (1939) 139-55.


Hrnenus, W.: Die rcimische Soldatensprache. Archiv fiir lateinische Lexicographie und Grammatikl2.2 (1902) 255-80.
Ffunoro, G.: Der Volksbegriff im Sprachschatz des Althochdeutschen und Altniederdeutschen. Junge Forschung. Untersuchun ger. zur Geistesgeschichte 8.
(Halle 1941).
HornuaNN, D.: Das spdtrcimische Bewegungsheer und die Notitia Dignitatum
(Epigraphische Studien 7.1-2). (Dusseldorf 1969 -70).
Hornrn, A.: Alt-Celtischer Sprachschatz. (Leipzigl896).
Horsrr, L. [Holstenius] (Ed.): Passio S. Bonifatii Romani. (Rome 1663).
Hor.ry H.C.: The Oxford Turkish-English Dictionary. Ed.3 rev. A.D. Andersory
Fahir Iz. (Oxford 1984).
Hunscnurrn, J.: Afr. 'tuivre"-dt. "Kocher": eine Wortefamilie hunnischen Ursprungs. In: Essais de Philologie Moderne. (Paris-Lidge 1953) 189-99.
Hurrsor, F.: Griechische und Rcimische Metrologie. Ed.'z(Berlin 1882).
Inrr.r, M.: Barditus. PW m.1 (1897) 10-11.
Iseac, B.: The Limits of Empire. The Roman Army in the East. Ed.'z (Oxford
1992).

]eNNr.nno

2004a

JauNrenn

2004b

Jrrnnrvs et

a1.1986

1932
Dnloscn1972

]oNus

Juo

1908

KaneNr-Ken,c.Nn

Krrrv
Krrrv

2008
1995

KsN{pr

1901

Krucs-Srrson
Kosr.rR.

1989

1961

1989

JeNNreno, S.: Armati, scutati et la cat6gorisation des troupes dans lAntiquit6


tardive. In: IlArm6e romaine de Diocl6tien ) Valentinien I" (Actes du Congrds
de Lyory 12-14sept.2002). Ed. Y. Le Bohes C. Wolff. (Paris 2004) 389-95.
J-a,NNr-r,nu, S.: Les formations tactiques en 6peron et en tenaille dans 1'arm6e
romaine. MEFR. Antiquit6 116.2 (2004) 1001-38.
]rrrnnvs, E.-Jrrrnnvs, M.-Scor1 R.-Cnoxr. B.: The Chronicle of |ohn Malalas
Byzantina Australiensia 4. (Melbourne 1986).
|ouns, CW.: Bede and Vegetius. CR46 (1932)248-9.

Dr ]oNcr,

P.: Philological and Historical Commentary on Ammianus Marcellinus XIX. (Groningen 1972).
]uo, J.: Travaux sur le Sarde (Comptes Rendus). Romania 37 (1908) 459-65.
Klruur, H.- KauaNr, R.: Cermanic Derivations of Romance Words. The |ournal of English and Romance Philology 60.3 (1961) 460-76.
Krrrv, G.: Ammianus Marcellinus: The Allusive Historian. (Cambridge 2008.)
Knrv, S.E.: Charters of St. Augustine's Abbey Canterbury and Minster-inThanet. (Oxford 1995).
Krurr, }.G.: Romanorum sermonis castrensis reliquiae collectae et illustratae.
Jahrbiicher fiir classische Philologie, suppl.26 (1901) 338-400
Krucr, F.- Srrnoro, E.: Etymologisches W<irterbuch der deutschen Sprache. 22.
Auflage. (Berlin 1989).
Korrnn, G.: Gotisches W<irterbuch. Ed.'?(Leiden 1989).

87

PHILIPRANCE

Korrns

1988

Korrns

1993

1985

Korr-lurz

Korras, T.G.: Byzantinishe Waffen. Ein Beitrag zur byza tinischen Waffenkunde von den Anfdngen bis zur lateinischen Eroberung. Byzantina Vindobonensia 17. (Vienna 1988).
Korras, T.: Tradition und Erneuerung im Friihbyzantinischen Reich am Beispiel der militdrischen Sprache und Terminologie. In: LArm6e romaine et
les barbares du IIIe au VIIe sidcles. Ed. M. Kazanski F. Vallet. M6moires de
lAssociation Frangaise dArch6ologie M6rovingienne 5. (Paris 1993) 39-44.
KoLLr,lvrz, A.: Das militdrwissenschaftliche Werk des sog. Maurikios. BY-

ZANTTAKA 5 (1985)
KoNnerov

1924

Kowelrwsxr

1844-9

1965

Kn,c.Hs

87-136.

KoNoerov, N.: Un ddtail des harnachements byzantins. In: M6langes offerts it


M. Gustave Schlumberger. (Paris 1924) lI 399 - 407

Kowerewsrr, J.E.: MoHrozr,cxo-Pyccro-Qpanqysxrri Czonapr/Dictionnaire


mongol-russe-frangais. (Kazan 1844-9).
KneHu, H.: Zt germ.-lat. Framea. Indogermanische Forschungen 70 (1965) 5052.

1987

Knr,rvmn, ].: Ein Grlizismus gotischer


Archiv u.r. 12 (1987) 197-207.

1992
1996

KneIrann, J.: Roga.

Knallrn

Knerrrn

1997

Knarvrxn
Kn,c.N{rn

Herkunft im Italienischen: bando.Balkan-

ZPE94 (1992) 185-90.

fiir fiinf germanische W<irter: agpaAaforcv,


pdv6ov, pouqbr.rv, pgarclov, oaQcbvtov. AfP 42 (1996) 113-26 (repr. Knaurn

Knannn, ].: Papyrusbelege


2011,39-53).

Knlrmn, ].: Von den Goten iiber die Griechen zu den Romanen:

punga, "Ta-

sche". In: Italica et Romanica. Festschrift fiir Max Pfister zum 65. Geburtstag.
Ed. G. Holtus, ]. Kramer, W. Schweickard. (Tubingen L997,1185-94 [repr. Kra-

2011
Knausn 1936

Knaurn

2004
KuuN 1972

Kudr"re

mer 2011, 269-771)


Knaurn, ].: Von Papyrologie zur Romanistik. (Berlin 2011).

Kuusr, W.: framea. In: Germanen und Indogermanen. Volkstum, Sprache,


Heimat, Kultur (Festschrift fiir Herman Hirt). Ed. H. Arntz. (Heidelberg 1935)
II s85-9.
Kudu4 W.: Crparerr,rxou Maapuru-s. Russ. trans. (St. Petersburg 2004.)
KuuN, H.: Das rcimische Kriegswesen in germanischen Wortschatz.ZfdAl0l
(1e72) 13-53.

L,rNrssnr

1902
1997

Larussnr

2003

Kur-axovsxu

Kurexovsrlt, l.A.: /pywn w,lpyarapuir. VizVrem 9 (1902)1-30.


Llnnrnl, P.-Y.: Lexique 6tymologique de I'irlandais ancien de |. Vendryes,
Lettre D. (Dublin-Paris 1997).
Llnrnrnl, P.-Y.: La langue gauloise. Description linguistique, commentaire

LaMNrsnr

1940
L,lNc 1885

d'inscriptions choisies. Ed.'? (Paris 2003).


Lanunnl, F.: Die Keil in der Taktik des Altertums. Gymnasium 51 (1940) 15-31.
LeNc, C. (Ed.): Vegetius: Epitoma rei militaris. Ed.'z (Leipzig 1885 [repr. Stutt-

LBG

Lexikon zur byzantinischen Grdzitdt besonders des

gart1957l).

LnsroyNsxy

Lnuorn
LnoNr

2001

1983

2003

9.-12.

fahrhunderts. Ed. E.

Trapp, W. Hcirandner et al.Yiennal01l.


LnnrovNsrv, I.: Armes et Guerriers barbares au temps des Grandes Invasion
(1\f au VI" siEcle aprds i.C.). (Paris 2001).
LrNoln, O.: Texte und Untersuchungen zum technischen Bereich der antiken
Poliorketik (Palingenesia 19). (Wiesbaden 1983).
LuoNr, B.: La Parafrasi Ambrosiana dello Strategicon di Maurizio: Uarte della
guerra a Bisanzio Bibliotheca Erudita Studi e Documenti di Storia e Filologia
22. (Milan 2003).

LswrN

2003
LrnNn-lnr 1977

Lnwrry A.: The Egyptian cunei. Tyche 18 (2003)73-6.


LruNueno, ].T.: Paulinus of Nola and Early Western Monasticism. (Cologne

1926-31
Lurto 1985

LrNrsnv, W.M. (Ed.): Glossaria Latina. (Paris 1926-31).


LuNo, A.A.: Neues zum mantischen Schlachtgesang der Germanen (Tac.

1e77).

LrNosnv

Germ.3,
LuNo

88

1988

1).

LuNo, A.A.

Maia 37 (1985) 263-70.


Cornelius Tacitus: Germania. (Heidelberg 1988).

(ed.): P.

THE PHILOLOGICAL EVIDENCE FOR GERMANI IN THE LATE ROMAN ARMY

LuNp

1991
1973

MesNcrrrN-HurreN

LuNo, A.A.: Kritischer Forschungsbericht zur "Germania" des Tacitus. ANRW


II 32.2 (1991) t989 :2222, 2341- 4.
MerNcsrN-Hnrrnry O.].: The World of the Huns: Studies in their History and

Culture. (London 1973).


Meuenas, N.: Die Musikinstrumente im byzantinischen Heer vom 5. bis zum
12. Jahrhundert. Eine Vorstellung der Quellen. IOB 51 (2001) 73-104.

Merrlus 2001
Ml'nrrN 1892
Mastnnrrr 1973

Mrnrrry

E.: Tufa.

ZfdA

(N.8. 24) 35 (1892) 96.

Mesrnurrr, C.A.: Ilelemento germanico nella toponomastica toscana dell'alto


medioevo. In: Atti del50 Congresso internazionale di studi sull'alto medioevo,
'Lucca e la Tuscia nell'alto medioevo' (Lucca 34 ott.1971). (Spoleto 1973) 64571.

2007
McCantNnv 1912
Marrsrws

Merrurws,

J.: The Roman Empire of Ammianus, rev. ed. (Baltimore 2007).


McCenrunv, E.S.: Figurative lJses of Animal Names in Latin and their Appli-

cation to Military Devices. (Lancaster, PA1912).

McMurrnlr

1963

McConnrlcr

1986

McCnrNors

1897

McMurmry R.: Soldier and Civilian in the Later Roman Empire. (Cambridge
Mass.1953.)

McGsrn

Murn

1991

McConnrcr; M.: Eternal Victory. Triumphal rulership in late antiquity,Byzantium, and the early medieval West. (Cambridge 1986).
McCnrNorr, ]W.: The Christian Topography of Cosmas, an Egyptian Monk.
(London 1897).
McGnrn, E.: Tradition and Reality in the Taktika of Nikephoros Ouranos.
DOP 4s (t991) 12e-40.
Mnrtn, H.: Spiitlat. sculca, "Wache, Wiichter" und seine romanischen und germanischen Verwandten. In: Lateinisch-romanische Etymologien. Ed. H. Mei-

1981

er. Untersuchungen zur Sprach- und Literaturgeschichte der romanischen


Vcilker X. (Wiesbaden 1981).
MnrNrxr, E.: Framea (91. Sprachliches). RGA2 IX (1995) 366-8.
Mrucrs, K.H.: Altaic Elements in the Proto-Bulgarian Inscriptions. Byzantion

Muxrrr 1995
MrNcts 1951

21 (1951) 8s-118.

Mruccr

1987

Msvrn-Ltirrr

1935

Mnuccr, R.: Lo strumento del bucinator. A. Surus e il cod. Pal. Lat. 909 di Vegezio. Bonner ]ahrbiicher 187 (1987) 259-72.
Mnvrn-Liimr, W.: Romanisches etymologisches Worterbuch. Ed.3 (Heidelberg
1e35).

Mrcr,ronrNI-Duvo
voN Mrxrosrcrr

1964

1885

Mtcrronrur, B.- Duno, A.: Prontuario etimologico della lingua ltaliana. Ed. a
(Turin 1964).
voN MrrrosrcH, F.: Etymologisches Wcirterbuch der slavischen Sprachen. (Vienna 1886).

MIrr.6.sscu

1968-9

MrHi.nscq H.: Les 6l6ments Latins des "Tactica-Strategica" de Maurice-Urbiciusetleur6choenn6o-grec. RESEE [I] 6 (1968) 481-98;UIl7 (1969) 155-66, [II!

Mrn,Esscu

1959

Mrnisscu

1959

Mrrri.rscu

1970

Mrrrisscu

1978

MrnXnscu, H.: Les termes de commandement militaires latins dans le Stratigicon de Maurice. Revue Roumaine de Linguistiqw 14 (1969) 261-72.
Mrni.rscu, H.: In margine alla mia recente edizione critica di Maurizio. In:
Studi filologici e storici in onore di Vittorio de Falco. (Naples 1971) 523-32
MrHi,uscq H. (Ed): Mauricius. Arta Militard (Scriptores byzantini vi). (Bucharest 1970).
MrHi.rscu, H.: La langue latine dans le sud-est de l'Europe. (Bucharest-Paris

Mtr.{rscu

1978-9

267-80.

1e78).

MIrNnn 1995
Monrrccnro 2002

Mrrri.nscu, H.: La litt6rature byzantine, source de connaissance du latin vulgaire. RESEE [U 16 (1978) t95-215;UIl17 (1979) 39-60, [IIU 359-83.
Mrrunn, N.P.: Vegetius: Epitome of Military Science. Ed.'z (Liverpool1996).
Monrrccnro, E.: got. bandwo. In: Lessico Etimologico Italiano, Germanismi 1,
fasc.2. (Wiesbaden 2002) 282-359.

89

PHILIPRANCE

Monr.rccnro

2003

Morrrccuro, E.: Presenze alloglotte nell'Italia dell'anno Mille. IJapporto delle


lingue germaniche al tipo italo-romanzo. In: Italia linguistica anno Mille Italia linguistica anno Duemila. Atti det XXXN Congresso internazionale di
Studi della Societi di Linguistica Italiana (Firenze 2000). Ed. N. Maraschio, T.

Monrrccuro

2004

1983
1993

Moscr Sassr

Moursos
Mucn

1897

Poggi Salani. (Rome 2003) 153-63.


Monrrccnro, E.: Die sprachliche Hinterlassenschaft der Germanen auf der

italienischen Halbinsel. In: Higermanry D.- Haubrichs, W'- Jarnut


Akkulturation. (Berlin 2004) 99 -110.

]'

@d'):

Moscr Sessr, M.G.: Il Sermo Castrensis. (Bologna 1983.)


Moursos, D.G.: Gothic puggs and Middle Greek nouyylov. Indogermanische
Forschungen 98 (1993) 219-30.
MucH, R.: Die Stiidte in der Germania des Ptolemdus. ZfdA (N.F. 29) 41, (1897)
97:143.

MucH, R.: Die Germania des Tacitus. Ed.3 rev. H. ]ankuhn, W. Lange. (Heidelbery1967).

Muctt7967
MtjlsNHorn

1890-1908

MiilmNHoEr, K.: Deutsche Altertumskunde. Ed.2 rev. M. Roediger. (Berlin


1890-1908).

1974

N.c.csl

Diktion der romischen Kommandosprache in den PraecePta des Pachomius. Zeitschrift fiir iigyptische Sprache und Altertumskunde l0l (1974)

Necrl,

P.:

t1420.

1927
N,cxsr 1915

N^p

Nscrrr

1918

Naa J.M: Ad Catonis librum de re militari. Mnemosyne 55 (1927)79-87.


Nrcxnr., G.: Hamalt fylkia. Beitrdge znr Geschichte der deutschen Sprache
und Literatur 40 (1915) 473-500.
Nrcxrr, G.: Hamalt fylkia und svinfylkia. Arkiv ftir Nordisk Filologi 34 (Ns 30)
(1918)284-34e.

NuriiorrN

2002

NruurNo

1965

NsuMeNN 1981
Nrcesrs 1998
NrnnnnrreNN
NoxosN

1905

1925

NnrEnxrry A.K.: Goths on Campaign from the Mid-Third to the Mid-Sixth


Centuries A. D. Fasciculi Archaeologiae Historicae 15 (2002) 9-15'
NnuuaNo, A.R.: Vegetius. PWRE suppl. X (1965)992-1020.
NrurueNry G.: Burg ($1. Sprachliches). RG,{, IV (1981) 117-118.
Nrclsrn, M.j.: Twilight of Empire: The Roman Army from the Reign of Diocle-

tian until the Battle of Adrianople. (Amsterdam 1998).


NrrounrvraNry M.: Contributions ir la critique et ir ltxplication des gloses latines. (NeuchAtel 1905).
Nononry E.: Die germanische Urgeschichte

zig-Berlin
OrtaN

1898
1995
lg1g-13

Or.runnrops

psosnssN

1940-45
p6nrz C.c.srno 2005
psNNrNcx

psnr

1990

pnn4Nrzor.4s

ptrz1977

1998

1959
poureNus1606
poronNv

poucnr

1955

psalrrs

1913

90

in Tacitus Germania.

Ed.3

(Leip-

1925.)

OneN, C.: A History of the Art of War. The Middle Ages from the Fourth to
the Fourteenth Century. (London 1898).
ONnnxro*s, A. (ed.): Vegetius: Epitoma rei militaris. (Stuttgart-Leipzig1995.)
PnornsrN, H.: Vergleichende Grammatik der keltischen Sprachen. (Gottingen
t909-13).
PrNrvrNcx, E.: r--origine hell6nique de burgus. Latomus 4 (1940-45) 5-21.

Pfnnz Cesrno, L.C.: Naturaleza y composici6n del sermo castrensis latino.


Em6rita 73.1 (2005) 73-96.
Prnr, G.: Tacitus' Germania. In: Griechische und Lateinische Quellen zur
Friihgeschichte Mitteleuropas bis zur Mitte des 1.. ]ahrtausends u. 2.. Ed. J.
Herrmann. Vol. 2. (Berlin 1990).
Pun.qnrzous, K.G. (Ed.): Ooiou'EQga[pr to0 ltgou (qya7. Thessaloniki 1998.
Pnrz, E.: Kamelaukion et mitra. Insignes byzantins imp6riaux et eccl6siastiques. Acta Universitatis Upsaliensis w.s. 15. (Stockholm 1977)'
Poronnv, J.: Indogermanisches Etymologisches W<irterbuch. (Bern 1959).
Popmuus, ].I.: Itinerarium Galliae Narbonensis... cui accedit Glossarium
Prisco-Gallicum seu ac Lingua Gallorum veteri. (Leiden 1606).
Poucu.r, P.: Die geheime Geschichte der Mongolen als Geschichtsquelle und
Literaturdenkmal. ieskoslovensk6 Akademie VEd, Archiv Orientdlni Supplementa IV. (Prague 1956).
Psarrns, S.B.: Grammatik der byzantinischen Chroniken. (Gottingen 1913)'

THE PHILOLOGICAL EVIDENCE FOR GERMANI IN THE LATE ROMAN ARMY


ReNcn

2004a

ReNcr

2004b

ReNcn

2010

R,q.Ncr

2014

RrNct

forthcoming

RaNcn, P.: Drungus, Aqooyyoq and Aqrouyytotl - a Gallicism and Continuity


in Roman Cavalry Tactics. Phoenix 58 (2004) 96-130.
RrNcr, P.:The Eulcum, the late Roman and Byzantine testudo: the Germanization of late Roman Tactics? GRBS 44 Q004) 265-326.
Rr.Ncr, P.: The De Militari Scientia or Miiller Fragment as a philological resource. Latin in the East Roman army and two new loanwords in Greek: palmarium
and*recala. Glotta 86 (2010) 63-92.
Rance, P.: Sculca, *sculcatoy exculcator and proculcator: the Scouts of the late Roman Army and a disputed Etymology. Latomus 73 Q014) 474-501.

ReNcn, P.: The Roman Art of War in Late Antiquity: The Strategikon of the
Emperor Maurice. A Translation with Commentary and Textual Studies. Bir-

mingham Byzantine, Ottoman and Modern Greek Monographs. (Farnham


forthcoming.)

Rtnvt

2004

RsrcHnr.{xnoN

Rnsrsnr

1984

vAN DER Ruus

Rrvrs

1961

1970

1999
1935

RorINsoN

RuIr.renr

1895-t987
1713

S,r'reuur

1963-4

on Ruccrsno

Sans,{H

Seofn

1978

1938

1929
1939
Scrsprrn 1664
S,c.NoEn

SeNonn

ScnrNx

1930

1970
1897
Scnol{rtro 1911
scnrrsacu
Snscx

ScHneIuu

Sornn

1954

1930

Sopnocr,ss

1887

SourrraxN-DrxoN

Sprrorr
Spuour

1975
2004a

1992

Rtnvr, M.D. (Ed.): Vegetius: Epitoma rei militaris. (Oxford 2004).


RxrcunNxnory G.: Zur rcimischen Kommandosprache bei byzantinischen
Schriftstellern. BZ 54 (L961) 18-27.
Rrsrnnr, G.: Goti, Tedeschi, Longobardi. Rapporti di Cultura e di Lingua Studi Grammaticali e Linguistici 16. (Brescia1984) 152-62.
vr.rv otn Rnnr, F.: Die germanischen W6rter in den langobardischen Gesetzen.
(Rotterdam 1970).
Rrvrs, J.B.: Tacitus. Germania. (Oxford 1999).
Ronrnsory R.P. (Ed.): The Germania of Tacitus. (Middletown, Conn. 1935 [repr.
Hildesheim 19911).
or Ruccrnxo, E.: Dizionario Epigrafico di Antichit) Romane. (Rome 1895-1987.)
RutNanr, T. (Ed.): Acta Primorum Martyrum Sincera et Selecta. Ed., (Amsterdam 1713. frepr. idem (ed.): Acta Martyrum. Regensburg 18591).
Slnanrtr, F.: Riflessi linguistici della dominazione longobarda nell'Italia mediana e meridionale. In: Atti e Memorie dellAccademia Toscana di Scienze e
Lettere La Colombaria 28 (N.s. 14) (1963-4) 123:249.
Sanrau, G.: La m6thode dAmmien Marcellin. Recherches sur la construction
du discours historique dans les Res Gestae. (Paris 1978).
Saofr, E.: Friihgermanische Wagenziige und Wagenburgen. Festschrift fiir
August Ox6. (Darmstadt 1938) 169J4.
SnNorn, E.: Frontin als Quelle fiir Vegetius. PhW 49 (1929) 1230-31.
Sanorn, E.: Die Germanisierung des rcimischen Heeres. HZ16O (lggg) 1-34.
Scnrrrrn, i. @d.): Arriani Tactica & Mauricii Artis militaris tibri duodecim.
(Uppsala 1564.)
ScHrNr; D.: Flavius Vegetius Renatus: Die Quellen der Epitoma rei militaris.
Klio Beiheft 22, N.t.9. (Leipzlg 1930 [repr. Nuremberg 1963]).
scurrmcH, E.: Byzantinische Metrologie. (Munich 1970).
Srncrg O.: Burgus. PWRE lll (1897) 1066-7
ScHoNrrro, M.: Worterbuch der altgermanischen Personen- und Vdlkernamen (Heidelberg 1911 [repr. Munich 1942]).
Scureuu, P.E.: Herrschaftszeichen und Staatssymbolik. Beitrdge zu ihrer Geschichte vom dritten bis zum sechzehnten Jahrhundert. Schriften der MGH
13. (Stuttgart 1954).
Sornn, j.: Lateinisches und Romanisches aus den Etymologiae des Isidorus
von Sevilla. (G<ittingen 1930).
Soruocrns, E.A.: Greek Lexicon of the Roman and Byzantine Periods from
B.C. 146 to A.D. 1100. (Cambridge, Mass. 1887 [repr. 1957, t97S)).
SouumnN, P.- DrxoN, K.R.: The Roman Cavalry from the First to the Third
Century AD. (London 1992).
Srrronr, M.P.: Vegetius (3.5) on Trumpets. AClass 18 (1975) 153-5.
Srrrnrr,, M.P.: Ancient Germanic Warriors. Warrior Styles from Trajan's Column to Icelandic Sagas. (London-New York 2004).

97

PHILIP RANCE

Sprrorr

2004b

SpsrNaeN

Snnrnrr, M.P.: The Four Earliest Auxilia Palatina. Revue des Etudes Militaires

Anciennes 1 (2004) 132-46.


Smnery H.: Glossarium Archaiologicum continens Latino-Barbaro, Peregrina, Obsoleta, etnovatae significationis vocabula. (London 1654).
SrnrNcnn, M.: Schlachtgesang. RGA2 XXVII (New York-Berlin 2002)122-4.
SrurNrn, A.: Byzantinisches im Wortschatz der Suda. Studien ztr byzantinischen Lexikographie. Ed. E. Trapp et al.Byzartina Vindobonensia 18. (Vienna

1.664

2002
1988

SprrNcrn
SrsrNsn

1988) 149-81.

2003
2007
SrrrnnNsoN-DrxoN 2002
Srrurn 1995
SrspHsNsoN
SrnpusNsoN

SrnrnrNsory P.: The Legend ofBasil the Bulgar-Slayer. (Cambridge 2003).


SrrnnuNsory I.P.: The Late Anglo-Saxon Army. (Stroud 2007.)
Srrnurwsory I.P.- DrxoN, K.: Roman Cavalry Equipment. (Stroud 2002).
Srnurn, H.: Framea ($2. Archiiologishes). RGA'z IX (New York-Berlin 1995) 3689.

SurrrvaN

2000

SwriNNr

2004

Surrrvary D.F.: Siegecraft: Two tenth-century instructional manuals by'Heron


of Byzantium'. Dumbarton Oaks Studies 36. (Washingtoru D.C. 2000.)
SvvANNn, I.: The Age of the Hippotoxotai. The Art of War in Roman Military
Revival and Disaster (491-636). Acta Universitatis Tamperensis 994. (Tampere
2004).

1981

SzLorczxv-Kenooss

1921

TsnnacrNr

2006

TrrnuNs-Gnossxopp

SzAorczxv-Kennosg S.: Der awarisch-tiirkische EinfluB auf die byzantinische


Kriegskunst um 600 (Anmerkungen zum Strategikon des Maurikios). In: K6ldy-Nagy, G. (Ed.): Turkic-Bulgarian-Hungarian Relations (VIth-XIth Centuries). Studia Turco-Hungarica 5. (Budapest198l) 63-71..
TnnnecrNr, B.: Gallico e Latino. A proposito di un recente libro sulla lingua
gallica. Rivista di Filologia e d'Istruzione Classica 49 (1921) 401-30.
TnsuNs-Gnossror,4 B.: Die vollstiindig erhaltene Leier des 6. |ahrhunderts aus
Grab 58 von Trossingeru Ldkr. Tuttlingen, Baden-Wiirttemberg. Germania 84
(2006)93-142.

1965
TrnrrNsecn 1973
THorr,rpsox

THoursory E.A.: The Early Germans. (Oxford 1965).

TnraNo.e.r,Hyruors

TrrruNr.rcn, H.: Studien zu Wortern volkssprachiger Herkunft in karolingischen Kcinigsurkunden. (Munich 1973).
TnreNoarnylrrnrs, M.A.: Die Lehnwtirter der mittelgriechischen Vulgdrlitera-

1909

tur. (Strasburg1909).
Tnout 1999
Ur.rnrrwooo 1999
Vrsv

2009

%.ror-HorueNN
Wsrscsnsnn

lg3L

vnN Wrnvnxn

1965

Tnour, D.E.: Paulinus of Nola: Life, Letters and Poems. Berkeley, CA,1999.
lJNornwoor, R.: Anglo-Saxon Weapons and Warfare. (Stroud 1999).
Yrs:v, Zs.: Praesidia et Burgi in the Early Roman Empire. In: Limes XX (XX
Congreso Internacional de Estudios sobre la Frontera Romana, Le6n, Septiembre 2006). Ed. A. MolliTlo et al.Anejos de Gladius 13. (Madrid 2009)989-96.

Warnr, A.-

Horrraeur.r, ].B.: Lateinisches Etymologisches Wcirterbuch. Ed.3


(Heidelberg 1938-54).
Wrrscnnnrn, L.: Die Sprache der Festlandkelten. In: Deutsches Archdologisches Institut, Romisch-germanische Kommission, XXter Bericht, 1930.

(Frankfurt-am-Main 1931) 147:226.


veN Wrnvsrn,H.: Burgus: versterking of nederzetting? (Avec un r6sum6 franEais: Burgus: fortification ou agglom6ration?) (Verhandelingen van de Koninklijke Vlaamse Academie voor Wetenschappen, Letteren en Schone Kuns-

1970
2004

WHeruoucH

Wunrrrn

Wnrrsv

1982

WHrrsy

1988

ten van Belgi6. Klasse der Letteren 59). (Brussels 1965).


WHanuoucu, j.: The Dialects of Ancient Gaul. (Cambridge, Mass. 1970.)
WHrrrun, E.L.: The Legion as Phalanx in the Late Empire [pt. I]. In: LArm6e
romaine de Diocl6tien ir Valentinien I"' (Actes du Congrds de Lyon, 12-14 sept.
2002). Ed. Y. Le Bohec, C. Wolff. (Paris 2004) 309-58.
WHrrrv, L.M.: Theophylact's Knowledge of Languages. Byzantion 52 (1982)
425-8.

Wtro

92

1970

Wnrrnv, L.M.: The Emperor Maurice and his HistoriarL Theophylact Simocatta on Persian and Balkan Warfare. (Oxford 1988).
Wrr.o, J.P.: Borrowed names for borrowed things? Antiquity 44 (1970) 125-30.

THE PHILOLOGICAL EVIDENCE FOR GERMANI IN THE LATE ROMAN ARMY

Wrusrtor

1909

Worsxe-Cortus196843
Wnson

1891

WrNs'rro1 E.O. (Ed.): The Christian Topography of Cosmas Indicopleustes.


(Cambridge 1909).
Worsxe-CoNus, W. (Ed.): Cosmas Indicopleustes: Topographie chr6tienne
(Sources chr6tiennes t4l, 159, 197). (Paris 1968-73).
Wnror, F.: Uber die Sprache der Ostgoten in Italien (Quellen und Forschungen
zur Sprach- und Kulturgeschichte der germanischen Vcilker 68). (Strasburg
1891).

ZtLLtrcus

1935

Zrorxowsxr

1999
2002

ZucxrnrvrlN

1995

Ztol"xowsrt

Zrrr"rrcus, H.: Zum Kampf der Weltsprachen im ostromischen Reich. (Helsin-

ki

1935).

J.: The Invention of the Tuba (Trurnpet). CW 92.4 (1999) 367-73.


Zrorxowsxr, ].: The Roman Bucina: A Distinct Musical Instrument? Historic
Brass Society Journal 14 (2002) 3l-58.
Zucrnnuary C.: Le Deureqrov pav6ov Kr.rvotavrtvtaKr^rv dans une 6pitaphe de

Zrolxowsxr,

Pylai. Tyche 10 (1995) 233-5.

93